Wnt signalosomes: What we know that we do not know

Abstract

Signaling through the Wnt/β‐catenin pathway is relayed through three multiprotein complexes: (1) the membrane‐associated signalosome, which includes the activated Wnt receptors, (2) the cytoplasmic destruction complex that regulates turnover of the transcriptional coactivator β‐catenin, and (3) the nuclear enhanceosome that mediates pathway‐specific transcription. Recent discoveries have revealed that Wnt receptor activities are tightly regulated to maintain proper tissue homeostasis and that aberrant receptor upregulation enhances Wnt signaling to drive tumorigenesis, highlighting the importance of signalosome control. These studies have focused on the detailed process by which Wnt ligands engage their coreceptors, LRP5/6 and Frizzled. However, the components that constitute the signalosome and the regulation of their assembly remain undefined. In this review, we discuss Wnt/β‐catenin signalosome composition and the mechanisms that regulate signalosome assembly, including the role of biomolecular condensates and ubiquitylation. We also summarize the evidence for the presence of Wnt ligand‐independent signalosome formation.

Wnt signaling occurs through a “signalosome” receptor complex, driven by oligomerization of the adaptor, Disheveled (Dvl). Both endocytosis and condensate formation have been proposed to mediate signalosome formation. Under pathological conditions, signalosomes may form in a Wnt ligand‐independent manner. Finally, signalosome assembly may be regulated by ubiquitylation of Wnt receptors.

INTRODUCTION

Wnt/β‐catenin signaling

The main effector of Wnt/β‐catenin (henceforth Wnt) signaling is the transcriptional coactivator, β‐catenin. In the absence of Wnt stimulation, β‐catenin is constitutively degraded by the cytosolic destruction complex. Axin acts as an essential scaffold for the destruction complex, in which the tumor suppressor, adenomatous polyposis coli (APC), also participates. Within the destruction complex, glycogen synthase kinase 3 (GSK3) and Casein kinase 1 alpha (CK1α) phosphorylate β‐catenin, marking it for recognition by the E3 ligase β‐TrCP, which ubiquitinates β‐catenin, targeting it for proteasomal degradation (Figure 1).^{[ 1} , ^{2 ]} Signaling is initiated when Wnt ligands engage the seven‐transmembrane receptor, Frizzled (Fz), and the single‐pass transmembrane coreceptor, LDL receptor‐related protein 6 (LRP6), or its paralog, LRP5. The ortholog of LRP5/6 in Drosophila is Arrow. In the current model, the cytoplasmic protein Disheveled (Dvl) is recruited to the membrane by Fz, and via its DIX domain bootstraps components of the Axin complex to the membrane, inhibiting the phosphorylation of the Axin‐bound β‐catenin. The proximity of LRP6 to Axin‐bound CK1 and GSK3 promotes LRP6 phosphorylation, which provides additional docking sites for further Axin recruitment to the plasma membrane in a positive feedback loop. As β‐catenin synthesis is relatively constant, the inhibition of β‐catenin phosphorylation via the direct interaction between phosphorylated LRP6 and Axin‐bound GSK3 allows for its cytoplasmic accumulation. Elevated cytoplasmic β‐catenin consequently translocates to the nucleus, where it binds TCF/LEF and other relevant nuclear factors to regulate downstream Wnt target genes (Figure 1).^{[ 1} , ^{2 ]}

FIGURE 1.

Current model of Wnt/β‐catenin signaling. In the absence of Wnt ligand, the transcriptional co‐activator, β‐catenin, is constitutively degraded, maintaining low levels of cytoplasmic β‐catenin (left). Wnt activation inhibits β‐catenin turnover, leading to its accumulation, nuclear translocation, and the transcriptional control of target genes (right).

Assembly and maturation of the Wnt receptor signalosome: The Dvl‐centric model

Wnt signaling is regulated by the dynamic assembly of large, higher‐order activated receptor complexes (termed the signalosome). The work of Cong et al. (2004) provided hints that the formation of receptor complexes is required for Wnt signaling, demonstrating that LRP6‐Fz receptor oligomerization promoted Wnt pathway activation.^{[ 3 ]} In a seminal study by the Niehrs lab, Bilic et al. (2007) first proposed a model for Wnt‐induced “LRP6 signalosome” formation.^{[ 4 ]} They reported that LRP6 punctate structures form below the plasma membrane shortly after Wnt treatment in HeLa and Xenopus embryo cells. These aggregates, resistant to solubilization by Triton X‐100, were observed via sucrose density gradient. Phosphorylated LRP6 was present in the denser fraction, which overlapped with large (ribosome‐sized) multiprotein complexes (referred to as “aggregates”) in Wnt‐stimulated cells. In contrast, little phosphorylated LRP6 was detected in samples from unstimulated cells. These studies involving detergent‐solubilized extracts suggested that LRP6 aggregates represent activated detergent‐resistant LRP6 protein complexes that also include the other Wnt pathway components Fz, Dvl, Axin, CK1, and GSK3.^{[ 4} , ^{5 ]}

Since the initial discovery of Wnt signalosomes by the Niehrs lab, emerging evidence indicated that the cytosolic adaptor protein Dvl may be a primary mediator of Wnt receptor aggregation and signalosome formation. Wnt ligands bind and activate the coreceptors LRP6 and Fz. Subsequently, activated Fz, via its KTXXXW C‐terminal motif, directly binds to the PDZ domain of Dvl, recruiting Dvl to the membrane.^{[ 6 ]} Wong et al. (2003) hypothesized that since the binding affinity between the PDZ domain of Dvl1 and the KTXXXW motif of Fz is relatively weak, other interacting regions between Fz and Dvl may facilitate their interactions.^{[ 6 ]} Subsequent studies indicated that the Dvl DEP domain can also mediate the binding of the intracellular transmembrane domain of Fz to Dvl.^{[ 7 ]}

Oligomerization of Dvl via its DIX and DEP domain was proposed to form a platform for LRP6 and Fz to copolymerize, resulting in high local Wnt receptor concentrations.^{[ 4} , ^{8 ]} Evidence for this mechanism comes from work by the Bienz lab.^{[ 9 ]} Based on structural and biophysical measurements, the Fz‐bound DEP domain of Dvl was proposed to undergo “domain swapping” once Wnt receptors are recruited into clathrin‐coated pits upon Wnt ligand binding.^{[ 9 ]} In this model, due to its high effective concentration in clathrin‐coated pits, the Dvl DEP domain dissociates from Fz and rearranges to form dimers. These dimers act as cross‐linkers for Dvl‐Axin head‐to‐tail polymers that are formed via interaction between the DIX domains in Axin and Dvl upon Wnt pathway activation.^{[ 10} , ^{11 ]} In addition to its role in Wnt receptor polymerization, Dvl has been proposed to recruit Axin and associated destruction complex components to the membrane via its DIX domain, where CK1 and GSK3 can phosphorylate LRP6. The phosphorylated PPPSPxS motifs on LRP6 were shown to directly bind to the catalytic pocket of GSK3, inhibiting its activity.^{[ 12} , ^{13 ]} Receptor clustering and signalosome formation have also been proposed to occur via the activity of the membrane protein TMEM59, by promoting intramembrane interactions between Fz and LRP6.^{[ 14 ]}

Membrane phosphatidylinositols have also been implicated in signalosome assembly and maturation. Dvl was shown to increase the membrane concentration of phosphatidylinositol (4,5)‐bisphosphate (PIP₂) by directly interacting with phosphatidylinositol kinase (PI4KII and PIP5KI). PIP₂ formation at the membrane was also shown to stimulate the phosphorylation of LRP6 at the Thr1479 and Ser1490 sites, which are indicative of receptor activation.^{[ 15 ]} Further evidence for the role of PIP₂ in signalosome formation comes from studies showing that PI(4,5)P₂ hydrolysis blocked the formation of active phosphorylated LRP6 in the heavy fraction pool.^{[ 15} , ^{16 ]} Finally, PIP5K was shown to promote Wnt receptor clustering via the action of the scaffold protein Daam2.^{[ 17 ]}

Dvl has also been proposed to play a role in non‐Wnt/β‐catenin signalosome‐like formation.^{[ 18 ]} Dvl polymerization and PI(4,5)P₂ production have been reported for Wnt5a signaling.^{[ 19} , ^{20 ]} Because PI(4,5)P₂ has been shown to recruit AP2 and clathrin to the plasma membrane, a similar mechanism involving PI(4,5)P₂‐induced signalosome formation may also occur in non‐Wnt/β‐catenin signaling.^{[ 5 ]} However, DEP domain swapping, which relies on the stable association between Fz and Dvl,^{[ 21 ]} is unlikely to occur. To date, evidence for a non‐Wnt/β‐catenin signalosome complex remains sparse, and further biochemical and cell biological evidence will be required to confirm its presence.

Based on these observations, a working model for Wnt signalosome assembly is as follows (Figure 2): (1) Wnt stimulation leads to the dimerization of Fz and LRP6 and subsequent recruitment of Dvl to the membrane via its direct interaction with Fz. (2) Dvl binds via its DIX domain to the DIX domain of Axin, recruiting a complex consisting of Axin and its two bound kinases, GSK3 and CK1, to the plasma membrane. (3) The subsequent high concentration of membrane‐localized Dvl results in Dvl self‐multimerization, creating a platform for further co‐polymerization of Fz and LRP6. (4) The mechanism that mediates the initial phosphorylation of the cytoplasmic tail of LRP6 by GSK3 remains unclear. Subsequently, LRP6 interacts with Axin^{[ 22} , ^{23 ]} via its phosphorylated PPPSPxS motifs,^{[ 24 ]} resulting in a positive feedback loop. (5) GSK3 and CK1 then phosphorylate the PPPSPxS motifs in cis and trans upon LRP6 multimerization.^{[ 8} , ^{25 ]} (6) Self‐aggregated Dvl at the plasma membrane also stimulates the formation of PIP₂, which recruits the Wnt receptor complex to clathrin‐coated pits^{[ 16 ]} or possibly other plasma membrane endocytic structures.^{[ 26} , ^{27 ]} These signalosomes serve as highly active centers to stabilize β‐catenin through inhibition of its phosphorylation by the destruction complex.^{[ 1} , ^{25 ]}

FIGURE 2.

Formation of Wnt‐induced signalosome. Wnt signalosome formation is proposed to occur through distinct steps.

Role of receptor internalization in Wnt signalosome formation and signaling

The role of endocytosis in Wnt signalosome formation has been reviewed in detail.^{[ 28} , ^{29 ]} Precedence for the role of endocytosis exists in other signal transduction pathways. In SMAD‐dependent TGF‐β signaling, pathway activation is initiated upon the formation of a plasma membrane receptor complex in clathrin‐coated pits. The complex is subsequently internalized into early endosomes, where its interaction with the adaptor proteins SARA and Dab2 propagates the TGF‐β signal.^{[ 30} , ^{31 ]} A similar mechanism has been proposed for Wnt signaling, in which Wnt, Fz, and LRP6 are internalized upon pathway activation.^{[ 32 ]} However, the mechanism by which Wnt ligands and receptors are internalized, and the requirement for this process has been controversial.^{[ 33 ]} Early in vivo studies suggested that endocytosis negatively regulates Wnt signaling; overexpression of Rab5, a protein involved in endosomal fusion, resulted in decreased Wingless target gene expression in the Drosophila wing disc.^{[ 34 ]} Conversely, evidence supporting a positive role for endocytosis in Wnt signaling came from a study demonstrating β‐catenin stabilization in murine L‐cells following inhibition of clathrin‐mediated endocytosis by use of chemical inhibitors and by knocking down clathrin.^{[ 35 ]} Furthermore, inhibition of dynamin, a GTPase that drives the fission of clathrin‐coated vesicles from the membrane, blocked Wingless internalization, resulting in its extracellular accumulation and reduction in the expression of Wingless target genes.^{[ 36 ]} Other evidence for an activating role of endocytosis comes from the work of Yamamoto et al. (2006) demonstrating Wnt pathway inhibition with nystatin treatment, which blocks caveolin‐mediated endocytosis.^{[ 27 ]} More general evidence for an endocytic mechanism in Wnt signaling activation comes from the discovery that the prorenin receptor, which is involved in the assembly of the vacuolar H+‐adenosine triphosphatase complex (V‐ATPase), a proton pump that acidifies intracellular compartments, links LRP6 with the V‐ATPase.^{[ 37 ]} Both the prorenin receptor and V‐ATPase were shown to be required for Wnt signaling.^{[ 38 ]} In addition, the μ2 subunit of the clathrin adaptor protein, AP2, was proposed to stabilize Dvl‐dependent Wnt signalosome assembly to propagate Wnt signaling.^{[ 40 ]} Finally, Saito‐Diaz et al. (2018)^{[ 38 ]} demonstrated that shifting cells to a lower temperature, a classic method to block endocytic pathways, inhibited Wnt‐stimulated stabilization of β‐catenin and Wnt signaling, which could be reversed by raising the temperature back to 37°C.

How endocytic mechanisms (clathrin versus caveolin‐mediated, or more recently, via multivesicular bodies [MVBs]) contribute to Wnt signalosome formation remains unresolved. Yamamoto et al. (2006) reported that LRP6‐GFP predominantly colocalizes with caveolin‐1 upon Wnt stimulation.^{[ 27 ]} Consistent with a role for caveolin in Wnt signalosome formation, Bilic et al. (2007) demonstrated that phosphorylated (activated) LRP6 aggregates overlapped with caveolin.^{[ 4 ]} The initial evidence that Wnt signalosome formation is a clathrin‐dependent process was based on the use of chemical clathrin inhibitors and siRNA knockdown of the clathrin heavy chain, which blocked Wnt3a‐mediated signaling.^{[ 35 ]} Subsequently, biochemical studies indicated that clathrin and the cargo adaptor protein AP2 were part of the LRP6 signalosome complex.^{[ 16 ]} Consistent with these biochemical studies, super‐resolution fluorescence microscopy showed the Wnt‐induced clustering of LRP6 and clathrin on the cell surface.

The controversy of whether caveolin or clathrin is required for receptor‐mediated endocytosis in Wnt signaling raises the possibility that the mechanism of receptor internalization may result in distinct downstream signaling outcomes. For example, Yamamoto et al. (2008) reported that Wnt3a induced LRP6 internalization via caveolin‐mediated endocytosis, whereas the secreted Wnt pathway inhibitor, DKK1, induced LRP6 internalization via clathrin‐mediated endocytosis, leading to stimulation and inhibition of the Wnt pathway, respectively.^{[ 41 ]} This phenomenon was also observed in TGF‐β signaling, in which receptor internalization by clathrin‐mediated endocytosis led to SMAD‐dependent signaling, whereas receptor internalization by caveolin‐mediated endocytosis led to SMAD7‐regulated receptor degradation.^{[ 30 ]} Wnt signalosome formation via clathrin or caveolin pathways was proposed to depend on cell type^{[ 5} , ^{16 ]} or developmental stage.^{[ 42 ]} Moreover, the clathrin and caveolin pathways may generate Wnt signalosomes with distinct properties. For example, studies by Saito‐Diaz et al. (2018)^{[ 38 ]} using chemical inhibitors and knockdown of clathrin and caveolin indicated that ligand‐independent Wnt receptor activation—upon loss of APC function—occurs via clathrin‐dependent endocytosis, whereas Wnt ligand‐mediated signaling occurs via the caveolin pathway in HEK293 cells. Whether signalosomes that form via the clathrin pathway exhibit different signaling properties from those formed via the caveolin pathway remains unclear. Notably, in the Saito‐Diaz study,^{[ 38 ]} several epithelial cell lines (e.g., HEK293, RKO, and RPE cells) mediated Wnt ligand‐dependent signaling via the caveolin pathway, whereas fibroblasts (e.g., L cells and MEFs), signaled via the clathrin pathway. The molecular details for how the clathrin versus the caveolin pathways internalize Wnt receptors remain to be more clearly defined. For example, Wnt‐induced phosphorylation of the cytoplasmic tail of LRP6 at Ser1579 by CK1 leads to clathrin‐mediated endocytosis, whereas blocking LRP6 Ser1579 phosphorylation results in its endocytosis via the caveolin pathway.^{[ 28} , ^{39 ]}

In addition to conflicting evidence as to whether clathrin‐ or caveolin‐mediated endocytosis mediates signalosome formation, clathrin and caveolin endocytosis were also found to be dispensable for Wnt signaling.^{[ 43 ]} Combining knockdown and knockout studies of clathrin and caveolin components with single‐cell imaging and multiple cell lines, Rim et al. (2020)^{[ 43 ]} found that disrupting clathrin and caveolin endocytosis did not block Wnt‐mediated β‐catenin accumulation or target gene transcription, suggesting that endocytosis is not required for Wnt signaling.^{[ 43 ]} However, since endocytosis is required for cell survival, it is possible that these knockouts were compensated by functional paralogs or that residual protein was sufficient to mediate receptor endocytosis in knockdown studies. Alternatively, Wnt signalosome formation in cells may occur via multiple endocytic pathways (e.g., clathrin and caveolin endocytosis and MBVs), and disruption of one pathway would be insufficient to significantly inhibit Wnt signaling.

The Wnt signalosome was also proposed to undergo sequestration in MVBs,^{[ 44 ]} which were similarly implicated in potentiating Toll/NF‐κB signaling in Drosophila.^{[ 45} , ^{46 ]} In this model, Wnt activation triggers the clustering of Wnt signalosomes into MVBs, resulting in global GSK3 inhibition. Supporting this mechanism, depleting HRS/Vps27, essential components of the ESCRT‐0 complex, blocked the Wnt‐mediated accumulation of β‐catenin.^{[ 26 ]} Furthermore, ESCRT components were recruited near the LRP6 receptor within 5 min of Wnt3a treatment.^{[ 33 ]} Additional evidence for the role of MVBs in Wnt signaling comes from studies showing that disruption of proteins that regulate the internalization of Wnt receptors into MVBs perturbs Wnt signaling.^{[ 47} , ^{48 ]} The transition between Wnt signalosome formation and its entry into MVBs may occur via macropinocytic cups of less than 200 nm, which later mature into MVBs.^{[ 29} , ^{33 ]} Although intriguing, the global changes in GSK3 activity as a mechanism of Wnt pathway activation in the MVB model have been difficult to measure and reconcile with GSK3's known roles in other signaling pathways.

To date, it is not clear if endocytosis simply represents a mechanism for receptor clustering and signalosome formation or also provides a vehicle to control receptor availability and/or activity. Perhaps endocytosis imparts different characteristics, including duration, amplitude, frequency, and feedback. Seen in this light, it is not difficult to envision the possibility that certain cell types are tied to a particular (clathrin, caveolin, or MVB) endocytic mechanism, such as in specific developmental stages, whereas other cell types may be more promiscuous and support Wnt signaling through multiple endocytic pathways. This possibility may underlie the conflicting conclusions that clathrin, caveolin, or neither are required for Wnt signaling. This idea also raises the intriguing possibility that the different endocytic mechanisms impart distinct pathway behavior that plays differing roles during development and adult tissue homeostasis.

Wnt ligand‐independent signalosome formation in APC mutant cells

Using cultured human cells and Drosophila, studies by Saito‐Diaz et al. (2018) showed that upon loss of APC, Wnt receptors are activated independently of Wnt ligand.^{[ 38 ]} This evidence supported the conclusion that Wnt receptor activation upon APC loss occurs via signalosome formation and requires clathrin‐mediated endocytosis. In this model (Figure 3), Wnt receptor clustering and internalization are mediated by the clathrin complex and are normally kept in check by APC via its binding to the clathrin adaptor, AP2, in the absence of Wnt ligand. In the APC‐deficient state, however, this brake on clathrin‐mediated receptor clustering and internalization is lost, and receptor clustering and signalosome formation occur in the absence of the Wnt ligand. Because LRP6 has been shown to bind to AP2 in signalosome formation,^{[ 16 ]} the APC function in inhibiting signalosome formation could be readily explained by competition between APC and LRP6 for binding to AP2 in a manner regulated by Wnt ligands. This new model for APC is controversial (summarized nicely by Zhong et al. (2021)^{[ 49 ]}) as Chen and He (2019)^{[ 50 ]} showed that CRISPR‐Cas9 mediated knockout of LRP5 and 6 in HEK293T clones in which APC was concomitantly knocked out failed to reduce β‐catenin levels or decrease Wnt signaling. In response to the work by Chen and He, single‐cell analyses were performed and demonstrated that individual APC‐depleted cells exhibited elevated β‐catenin levels, which were lowered upon LRP6 silencing.^{[ 51 ]}

FIGURE 3.

Proposed mechanism of Wnt ligand‐independent signalosome formation. Adenomatous polyposis coli (APC), through its binding to the cargo component, AP2, in the clathrin complex, acts as a brake to block the recruitment of Wnt receptors into clathrin‐coated pits in the absence of Wnt ligands (left). Loss of APC results in clustering of Wnt receptors into clathrin‐coated pits, promoting the formation of the Wnt signalosome (right).

Further evidence for the role of APC in receptor activation was provided by proximity labeling experiments with an LRP6‐APEX chimera, which showed that APC is enriched, followed by strong inhibition within minutes of Wnt stimulation, suggesting a possible removal of APC from the receptor complex.^{[ 33 ]} Notably, restoration of APC in APC‐mutant SW480 colorectal cancer cells blocked macropinocytosis, a process proposed to promote signalosome formation.^{[ 52 ]} Consistent with the role of APC in signalosome formation, Wnt receptor activation was required for high levels of Wnt signaling in colorectal cancer cells.^{[ 53 ]} In addition, LRP6 and its activated, phosphorylated form have been shown to be upregulated in many human colorectal cancers,^{[ 54} , ^{55 ]} and phospho‐LRP6 is indicative of a worse prognosis.^{[ 55 ]} Two recent studies, however, failed to demonstrate that the knockdown of LRP6 inhibited Wnt signaling in APC mutant cells.^{[ 56} , ^{57 ]} Resolution of the role of APC in receptor signaling awaits experimental evidence describing a detailed molecular mechanism for how APC directly regulates Wnt receptor signaling. Finally, as an additional twist to the story of the role of APC in Wnt signaling, a recent study suggests that APC facilitates the translocation of the β‐catenin destruction complex to the cell membrane to engage with Wnt receptors.^{[ 58 ]}

Regulation of Wnt receptors in the signalosome by ubiquitylation

Ubiquitylation plays a critical role in regulating key cell functions that include cell cycle, DNA repair, and signal transduction.^{[ 59 ]} The role of ubiquitylation in β‐catenin degradation has been intensely studied since the discovery that control of β‐catenin turnover is the main mechanism by which signaling is regulated and that the vast number of cancer mutations in the pathway aberrantly stabilize β‐catenin. Recently, there has been great interest in the role that ubiquitylation plays in Wnt receptor homeostasis. The control of LRP6 and Fz levels in vertebrates occurs via their ubiquitylation by the E3 ligases, RNF43 and ZNRF3, which target the Wnt receptors for degradation through the lysosomal pathway.^{[ 60 ]} Mutations in RNF43 have been found in human cancers, and conditional loss of RNF43 and ZNRF3 in mouse models led to hyperactivation of the Wnt pathway and hyperplasia.^{[ 61} , ⁶² , ^{63 ]} RNF43 and ZNRF3 themselves are regulated by the secreted R‐spondins, which at times in complex with the LGR4/5/6 family of seven transmembrane receptors inhibit RNF43 and ZNRF3 E3 ligase activities (Figure 4).^{[ 64 ]}

FIGURE 4.

Two proposed roles for ubiquitylation in regulating Wnt receptor activity. Ubiquitylation of Wnt receptors can lead to their internalization and lysosomal degradation (left). Alternatively, ubiquitylation can sterically block Wnt receptor clustering and signalosome formation (right).

Several deubiquitinating enzymes (DUBs) were also found to regulate Wnt receptors. USP8 and USP6 constitutively deubiquitylate Fz,^{[ 65} , ⁶⁶ , ⁶⁷ , ^{68 ]} whereas USP19 deubiquitylates LRP6 in the endoplasmic reticulum, facilitating its transport.^{[ 69 ]} Another DUB, USP42, prevents the ubiquitin‐mediated turnover of RNF43 and ZNRF3 to promote Wnt receptor degradation.^{[ 70 ]} Recent work identified the USP46 complex as a DUB that catalyzes the deubiquitylation of LRP6 in human cultured cells and Drosophila.^{[ 71} , ^{72 ]} Upon Wnt ligand activation, the USP46 complex is recruited to and deubiquitylates LRP6, stabilizing its activity (Figure 4).^{[ 71} , ^{72 ]} Interestingly, although the USP46 complex is evolutionarily conserved in vertebrates and Drosophila, there is no homolog for RNF43 or ZNRF3 in insects, suggesting the presence of a yet‐unidentified evolutionarily conserved E3 ligase that targets the Wnt receptors.

Studies on the ubiquitylation of Wnt receptors have focused on their ubiquitin‐mediated degradation. The covalent modification of proteins by ubiquitylation, however, has been shown to affect their conformation as well as their interactions with other proteins.^{[ 59 ]} One example of the latter is demonstrated by the proteasomal shuttle factor UBQLN2, which undergoes liquid–liquid phase separation (LLPS) into protein‐rich drops with multivalent, weak protein:protein interactions.^{[ 73 ]} Polyubiquitylation of UBQLN2 disrupts its oligomerization, blocking its phase separation behavior (see the following section: The role of biomolecular condensate in Wnt signalosome formation). In the Wnt pathway, ubiquitylation of DVL's DIX domain was shown to block its oligomerization, a mechanism thought to be necessary for signalosome formation.^{[ 74 ]} Similarly, studies by Ng et al. (2023) suggest that ubiquitylation of LRP6 not only promotes receptor degradation but also inhibits the oligomerization and assembly of Wnt signalosomes (Figure 4).^{[ 71 ]} This conclusion is based on the observation that the decrease in Wnt signaling appeared significantly greater than the observed decrease in the level of LRP6 upon knockdown or knockout of the USP46 complex in cultured human cells and Drosophila.^{[ 71} , ^{72 ]} These results are consistent with the initial studies on the effects of RNF43 and ZNRF3 on Wnt signaling, in which there was discordance between the small change in Wnt receptor levels and a much greater change in signaling output.^{[ 60 ]}

The role of biomolecular condensate in Wnt signalosome formation

Recently, the Wnt signalosome and the destruction complex were proposed to exist as biomolecular condensates.^{[ 13} , ⁷⁵ , ^{76 ]} Biomolecular condensates are membraneless compartments in the cell that undergo self‐assembly to form localized regions of concentrated components. These condensates most often form via reversible LLPS^{[ 76 ]} between a dense and a more dilute phase of the bulk cellular environment. The physiochemical properties of LLPS have been proposed to allow for the control of enzymatic reactions, localize and sequester proteins, and mediate cellular signaling.^{[ 76 ]}

Several lines of evidence suggest that both the β‐catenin destruction complex and signalosome undergo LLPS to form biomolecular condensates. The destruction complex scaffold proteins, Axin and APC, exhibit features found in proteins involved in condensate formation: regions allowing for oligomerization, multiple protein binding sites, and extended intrinsically disordered regions.^{[ 75 ]} Experimental evidence supporting destruction complex condensate formation comes primarily from overexpression studies of Axin at either supra‐ or near‐physiological levels in cultured cells^{[ 77} , ⁷⁸ , ⁷⁹ , ^{80 ]} and Drosophila embryos,^{[ 75 ]} in which the size of puncta visible by microscopy is increased upon overexpression of APC.^{[ 81} , ^{82 ]} Despite attempts in these studies to limit the degree of overexpression to near endogenous levels, the possibility of artifacts, including improper folding, still exists. Definitive evidence for the presence of destruction complex condensates will, therefore, require the knock‐in of reporters into the endogenous gene loci.

The early studies of Bilic et al. (2007)^{[ 4 ]} indicated that Wnt ligands trigger the formation of membrane‐associated phospho‐LRP6 aggregates that also contain Dvl, Axin, and GSK3. As their general properties of receptor clustering (including receptor oligomerization), association with cytoplasmic complexes, and multivalent protein interactions are similar to systems that undergo LLPS,^{[ 83} , ^{84 ]} these LRP6 signalosomes were proposed to represent condensates. Experimental evidence for LLPS in signalosome formation has focused primarily on Dvl due to its central role in receptor aggregation and capacity to undergo head‐to‐tail polymerization, a proposed feature of biomolecular condensates.^{[ 85 ]} Whereas previous overexpression studies resulted in the presence of Dvl puncta, whether these puncta can indeed form at endogenous concentrations remains an unresolved question. Rigorous studies by Ma et al. that coupled CRISPR‐mediated knock‐in of GFP into the endogenous Dvl2 locus with TIRF analysis revealed that Wnt stimulation induced the membrane localization of Dvl2 and the formation of oligomers; however, these contained no more than five Dvl molecules per complex.^{[ 86 ]} In contrast, Kang et al. detected relatively large puncta of 0.2–0.5 µm when endogenous Dvl2 was tagged with mEGFP (into which Axin was recruited) and were enhanced with Wnt ligand stimulation.^{[ 81 ]} The differences between these two findings can be reconciled by studies from Schubert et al. (2022)^{[ 87 ]} and Kan et al. (2022),^{[ 88 ]} which showed that when endogenous Dvl2 was tagged with mEos3.2 or GFP, respectively, puncta were present in only a subpopulation of cells (20%). Significantly, in the Schubert et al. study,^{[ 87 ]} these puncta were associated with interphase centrosomes in a Wnt‐dependent manner and were dispensable for Wnt signaling. This finding is consistent with a previous study by Smalley et al. (2005),^{[ 89 ]} which showed that expression of Dvl2 at low levels did not induce puncta formation but nonetheless activated Wnt signaling. Clearly, additional studies are needed to determine the biological significance of these Dvl2 condensates and their role in Wnt signalosome formation and signaling.

CONCLUSION

Initially identified as a biochemical and microscopic phenomenon associated with activated Wnt receptors, the signalosome has become an important conceptual framework for studying the activation of the Wnt pathway at the cell surface. The exact composition of the signalosome remains unknown; accumulating evidence indicates a minimal core complex of Wnt ligand, phosphorylated LRP6, Fz, Dvl, Axin, GSK3, and CK1. Although components of the clathrin complex, including AP2, are associated with the Wnt signalosome, they represent factors necessary for signalosome assembly rather than bona fide Wnt pathway components that mediate signaling. Finally, the importance of signalosome regulation is suggested by the fact that signalosomes can also form under pathological conditions, such as upon loss of APC function. Intense effort has been made to study the regulation of Wnt receptor turnover by the ubiquitin‐proteasome pathway. Given the moderate effects on Wnt receptor levels, it is intriguing that the main impact of Wnt receptor ubiquitylation may be to disrupt signalosome formation. This type of regulation would be consistent with the notion that the Wnt signalosome is a dynamic structure that provides a major hub of Wnt pathway regulation in time and space, possibly achieved via biomolecular condensate formation.

AUTHOR CONTRIBUTIONS

Heather Hartmann and Ghalia Saad Siddiqui wrote the initial manuscript with input from Jamal Bryant. Subsequent revisions were made by David J. Robbins, Vivian L. Weiss, Yashi Ahmed, and Ethan Lee. All the authors approved the initial and revised versions.

CONFLICT OF INTEREST STATEMENT

Ethan Lee and David J. Robbins are co‐founders of StemSynergy Therapeutics, a company that seeks to develop inhibitors of major signaling pathways (including the Wnt pathway) for the treatment of cancer. The remaining authors declare that they have no competing interests.

Hartmann, H. , Siddiqui, G. S. , Bryant, J. , Robbins, D. J. , Weiss, V. L. , Ahmed, Y. , & Lee, E. (2025). Wnt signalosomes: what we know that we do not know. BioEssays, 47, e2400110. 10.1002/bies.202400110

DATA AVAILABILITY STATEMENT

Data sharing is not applicable to this article as no new data were created or analyzed in this study.