Induction of Gsk3β-β-TrCP Interaction Is Required for Late Phase Stabilization of β-Catenin in Canonical Wnt Signaling*

Background: Despite being a key component of the Wnt pathway, how Gsk3β is regulated in Wnt signaling remains elusive. Results: Wnt simulation induces Gsk3β-β-TrCP interaction and monoubiquitination of Gsk3β, leading to inhibition of β-catenin recruitment of β-TrCP. Conclusion: β-Catenin stabilization in late Wnt signaling requires increased Gsk3β-β-TrCP interaction. Significance: First evidence that Gsk3β is regulated by monoubiquitination in Wnt signaling. A pivotal step in canonical Wnt signaling is Wnt-induced β-catenin stabilization. In the absence of Wnt, β-catenin is targeted for β-transducin repeats-containing proteins (β-TrCP)-mediated degradation due to phosphorylation by glycogen synthase kinase 3 (Gsk3). How canonical Wnt signaling regulates Gsk3 to inhibit β-catenin proteolysis remains largely elusive. This study reveals novel key molecular events in Wnt signaling: induction of Gsk3β ubiquitination and Gsk3β-β-TrCP binding. We found that Wnt stimulation induced prolonged monoubiquitination of Gsk3β and Gsk3β-β-TrCP interaction. Monoubiquitination did not cause Gsk3β degradation nor affects its enzymatic activity. Rather, increased monoubiquitination of Gsk3β/Gsk3β-β-TrCP association suppressed β-catenin recruitment of β-TrCP, leading to long-term inhibition of β-catenin ubiquitination and degradation.

Many different models have been proposed to explain how Wnt signaling inhibits ␤-catenin degradation including: 1) Wnt induces rapid disruption of Axin/Gsk3 interactions, which diminishes ␤-catenin phosphorylation and causes initial stabilization of ␤-catenin (6). 2) Wnt abrogates ␤-TrCP recruitment to ␤-catenin and blocks ␤-catenin ubiquitination within the destruction complex (7). 3) Wnt induces membrane sequestration of Axin1/Gsk3 complex by binding Wnt co-receptor LRP5/6 (the low-density lipoprotein receptor-related protein 5/6) (8 -10). 4) Wnt signaling inhibits Gsk3 phosphorylation of ␤-catenin via phosphorylated PPPSPXS motifs of Wnt coreceptor LRP6 (11,12). 5) Wnt induces Axin dephosphorylation by protein phosphatase PP1 within the LRP6/Axin signaling complex. Dephosphoryaltion of Axin changes its conformation, leading to disassembly of destruction and signaling complexes (8). 6) Wnt induces Axin1 degradation, which contributes to chronic Wnt signaling (6,13). 7) Wnt-induced sequestration of Gsk3 from the cytosol into multivesicular bodies (14) is also considered as possible mechanisms for long-term Wnt signaling, although a later study failed to confirm it (7). Overall, these models and others that are not elaborated here provide some mechanistic explanations for Wnt-induced stabilization of ␤-catenin; however, despite its pivotal role in the Wnt pathway, how Gsk3 is regulated by Wnt to achieve ␤-catenin stabilization remains a fundamental question in the field of Wnt signaling.
Post-translational modifications (PTMs) play central roles in creating a highly dynamic relay system that reads and responds to intracellular or environmental changes, by reversibly regulate protein functions, such as activity, stability, localization, and protein-protein interaction without requiring de novo protein synthesis (15). This study has investigated Wnt-induced ␤-catenin stabilization from an important but never explored angle-Gsk3␤ ubiquitination. We find that Wnt induces prolonged multi-monoubiquitination of Gsk3␤ and Gsk3␤-␤-TrCP binding. Induction of Gsk3␤ ubiquitination/Gsk3␤-␤-TrCP interaction inhibits ␤-catenin recruitment of ␤-TrCP, leading to inhibition of ␤-catenin ubiquitination. Together, our results have identified the induction of Gsk3␤ ubiquitination/ Gsk3␤-␤-TrCP interaction as novel key molecular steps required for Wnt-induced chronic stabilization of ␤-catenin.

EXPERIMENTAL PROCEDURES
Plasmids-The expression plasmids of hemagglutinin (HA)-Gsk3␤ and VSVG-LRP6 were obtained from Addgene. Plasmid expressing HA-tagged ␤-TrCP has been described previously (16). Gsk3␤ mutants containing lysine to arginine mutations were generated by PCR-directed mutagenesis. The Ubiquitin-Gsk3␤ fusion constructs were generated by fusion of ubiquitin to the N-terminal of Gsk3␤ (without first methionine) via a two-amino acid linker (Glu-Phe). To prevent ubiquitin fusion proteins function as ubiquitin or ubiquitin-like modifiers in the cells, the C-terminal diglycine of ubiquitin were deleted (⌬GG) in the fusion constructs. pcDNA3-Flag-ubiquitin was generated by inserting the ubiquitin coding sequence into pcDNA3-Flag, which was kindly provided by Dr. Jens Lykke-Andersen (University of California, San Diego). The sequence of shRNA oligonucleotides for Gsk3␤ (sense: 5Ј-GGACAAGAGATTTAAGAAT-3Ј) was reported previously (17). The annealed primers were ligated into pLKO.1 which was obtained from Addgene to construct lentiviral-based vector for Gsk3␤ knockdown. Four synonymous mutations were introduced into the shRNA targeting sequence of Gsk3␤ (sense: 5Ј-GGATAAGAGGTTTAAAAAC-3Ј) to generate shRNA-resistant Gsk3␤. Plasmids for shRNA-resistant Gsk3␤, and its mutants were generated by PCR-directed mutagenesis. All plasmid constructs were verified by DNA sequencing. pLKO.1-TRC control was obtained from Addgene. The mouse Gsk3␣-specific shRNA in pLKO.1-puro was purchased from Sigma. siRNA oligonucleotides for ␤-TrCP1/2 have been described previously (18) and were purchased from Thermo Scientific. Control siRNA oligos were purchased from Santa Cruz Biotechnology.
Cell Culture and Production of Wnt3a-conditioned Medium-Human embryonic kidney (HEK) 293T cells and human colon cancer cell line SW480 cells were obtained from ATCC. Immortalized Gsk3␤ ϩ/ϩ and Gsk3␤ Ϫ/Ϫ mouse embryonic fibroblasts (MEFs) were generously provided by James Woodgett, Ontario Cancer Institute, Canada. These cells were grown in DMEM supplemented with 5% fetal bovine serum (FBS) in a 37°C humidified incubator containing 5% CO 2 . Wnt3a-producing L cells and control L cells were obtained from ATCC and used for generating Wnt3aconditioned medium (Wnt3a-CM) and control-conditioned medium (control-CM) according to ATCC's instructions. In experiments involving Wnt stimulation, to ensure that cells exert a maximal response to Wnt, cells were maintained at about 40% (for MEFs) to 70% (HEK293T) confluent state before Wnt3a treatment.
Transient Transfection, RNA Interference, Lentivirus Production, and Infection-Plasmid and siRNA transient transfections were performed using PolyJet In Vitro DNA Transfection Reagent (SignaGen) according to the manufacturer's instruction. The transfection efficiency for MEFs was about 60 -70%. Gsk3␤ shRNA lentiviral particles were produced in HEK293T cells by transfection of the lentiviral vector expressing shRNA against Gsk3␤ with the third generation packaging systems (Addgene). The media containing viral particles were filtered through syringe filters and subsequently used to infect target cells. Cell lines stably expressing Gsk3␤ shRNA were established by puromycin selection.
Immunoprecipitation-Cells were washed with cold PBS twice and then lysed with M-PER buffer (Thermo) supplemented with protease inhibitor and 20 mM N-ethylmaleimide (NEM) (Sigma). The whole cell lysates were collected by centrifugation and pre-cleared with Protein G-Sepharose beads at 4°C for 30 min. The cleared lysates were incubated with indicated antibody together with Protein G-Sepharose beads at 4°C for 3-4 h. The immunoprecipitates were washed three times with IP lysis buffer containing 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, and 10% glycerol, separated by SDS-PAGE, and then immunoblotted with indicated antibodies. Input corresponded to 5% of the total lysates used for the IPs and was verified by immunoblotting.
In Vivo Ubiquitination Assay-Cells were washed with cold PBS and then lysed with IP lysis buffer containing protease inhibitor, 20 mM NEM and 1% SDS. 1% SDS was added to disassociate proteins complexed with Gsk3␤. The lysates were then diluted with IP lysis buffer to bring the final SDS concentration to 0.1% and subjected to pre-clear with Protein G-Sepharose beads at 4°C for 30 min. The cleared lysates were incubated with anti-Gsk3␤, or EZview Red anti-HA, or anti-FLAG M2 affinity gel (Sigma) at 4°C for 3-4 h. The immunoprecipitates were washed three times with RIPA buffer (Thermo), separated by SDS-PAGE and then immunoblotted with indicated antibodies.
In Vitro Kinase Assay-Briefly, HEK293T cells were transfected with empty vector or plasmids expressing HA-tagged GSK3␤ or its mutants. Twenty-four hours after transfection, the cells were lysed with lysis buffer. The cell lysates were incubated with EZview Red anti-HA affinity gel (Sigma) at 4°C for 3 h. The bead-bound immunoprecipitates were washed with lysis buffer for three times followed by a final kinase buffer wash, then incubated with full-length recombinant human CK1␣1 (5 ng, SignalChem), recombinant full-length human ␤-catenin (200 ng, Abcam), 10 mM ATP, kinase buffer (Cell Signaling) in a total volume of 15 l at 37°C for 2 h. The reaction product was subjected to SDS-PAGE. Kinase activity was evaluated by immunoblotting with antibody recognizing phosphorylated ␤-catenin at Ser-33, Ser-37, and Thr-41.
In Vitro Binding Assay-HA-tagged Gsk3␤ and its mutants were generated using the TNT T7 quick-coupled transcription/ translation kit (Promega). 15 l out of 25 l of the translation product was incubated with EZview Red anti-HA affinity gel (Sigma) for 1 h at 4°C in M-PER buffer. The immunoprecipitates were extensively washed with lysis buffer and then incubated with 750 ng GST-␤-TrCP (Novus) in binding buffer containing 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1% Nonidet P-40, and 10% glycerol for 2 h. After washing with lysis buffer, beads were analyzed by SDS-PAGE, followed by immunoblotting.
Protein Stability Assay-HEK293T cells were transfected with vectors expressing wild type (wt) HA-GSK3␤ or mutant proteins. 24 h after transfection, the cells were treated with or without 20 g/ml cycloheximide (CHX) in control-CM or Wnt3a-CM for indicated time. Cells were harvested after CHX treatment, and the whole cell lysates were used for immunoblotting.
In Vitro Ubiquitination Assay of Gsk3␤-Ubiquitination experiments were carried out in 1ϫ ubiquitination assay buffer (25 mM Tris-Cl at pH 7.5, 5 mM MgCl 2 , 100 mM NaCl, 2 mM ATP, 1 mM dithiothreitol) at a final volume of 20 l containing: 100 ng of GST-Gsk3␤ (BPS Bioscience), 130 ng of E1 (Boston Biochem), 230 ng of E2 (UbcH5b from Boston Biochem, UbcH5a, and UbcH7 from LifeSensors), 300 ng of SCF ␤-TrCP complex consisting of Rbx1, Cul1, Skp1, and ␤-TrCP (Millipore), 3 g of FLAG-ubiquitin or methylated ubiquitin (Boston Biochem). The reaction mixture was incubated at 37°C for 1 h, and then subjected to SDS-PAGE followed by immunoblotting with indicated antibody. WT HA-Gsk3␤ and HA-Gsk3␤-KKKK15/27/32/36RRRR mutant proteins were synthesized using the TNT T7 quick-coupled transcription/translation kit (Promega). To purify in vitro translated HA-Gsk3␤ proteins, 5 l out of 25 l of the translation product were used for immunoprecipitation with EZview Red anti-HA affinity gel (Sigma). The immunoprecipitates were subjected to in vitro ubiquitination assay as described above. All experiments were performed independently at least three times.

RESULTS AND DISCUSSION
Wnt Stimulation Induces Ubiquitination of Gsk3␤-Gsk3␤ is an ubiquitination substrate (19). In HEK293T cells, the basal level of Gsk3␤ ubiquitination was low since ubiquitination of HA-Gsk3␤ could be detected by immunoblotting only when ubiquitin was overexpressed (Fig. 1A). Treatment of the Con8 rat mammary epithelial tumor cell with the dexamethasone induces ubiquitination of Gsk3␤ (19); we therefore tested whether Wnt stimulation could induce Gsk3␤ ubiquitination. Similar to the case of ectopically expressed HA-Gsk3␤, the basal ubiquitination level of endogenous Gsk3␤ in HEK293T cells was barely detectable (Fig. 1B). Treatment of the cells with Wnt3a-CM, however, remarkably increased the level of ubiquitinated Gsk3␤ in a time course-dependent manner (Fig. 1B). Gsk3␤ is a shared pathway component, and only Axin-bound Gsk3 (Ͻ10% of cellular Gsk3) is engaged in the canonical Wnt pathway (20,21). Based on this, we expected that less than 10% of cellular Gsk3␤ were ubiquitinated upon Wnt stimulation. The result of immunoblotting on immunoprecipitated samples indeed confirmed that only a small portion of cellular Gsk3␤ proteins were ubiquitinated (Fig. 1C).
Ubiquitination often triggers proteasomal degradation of the substrate. The accumulation of ubiquitinated Gsk3␤ upon Wnt stimulation could result from (1) Wnt-induced stabilization of ubiquitinated Gsk3␤ or (2) induction of Gsk3␤ ubiquitination. To discriminate between these two possibilities, we compared the levels of ubiquitinated Gsk3␤ in the absence and presence of proteasome inhibitor MG132. Treatment with MG132 did not affect the levels of total and ubiquitinated Gsk3␤ under Wnt "on" and "off" conditions ( Fig. 1D), hence supporting the second possibility: induction of Gsk3␤ ubiquitination.
It appeared that most of the PTM found in endogenous Gsk3␤ was monoubiquitination, whereas multi-monoubiquitination of Gsk3␤ occurred when manipulating the system in vitro or via overexpression of ubiquitin. To validate monoubiquitination of Gsk3␤, we examined whether using wild type ubiquitin or methylated ubiquitin exerts the same or distinct ubiquitination patterns of Gsk3␤. The in vitro ubiquitination assay result showed that using Flag-ubiquitin (9.5 kDa) and methylated ubiquitin (8.5 kDa) caused a slight difference in band size in lane 2 and lane 3, but the overall Gsk3␤ ubiquitination patterns in these two lanes were very similar (Fig. 2E), hence confirming monoubiquitination of Gsk3␤. The size of the highest ubiquitinated bands further suggested that about six lysines within Gsk3␤ were modified by single ubiquitin moieties (Fig. 2E), which could explain why mutation of four lysines on Gsk3␤ substantially, but not completely, abrogated in vivo monoubiquitination of Gsk3␤ (Fig. 1E). Interestingly, mutation of lysines 15/27/32/36 appeared to be sufficient to abrogate in vitro ubiquitination of Gsk3␤ (Fig. 2F). This seemly discrepancy between in vivo and in vitro might result from different ubiquitination efficiencies in different ubiquitination systems.
Ubiquitination of Gsk3␤ Does Not Alter Gsk3␤ Stability and Enzymatic Activity-Protein functions are often regulated by PTMs (15). We examined whether Gsk3␤ stability and activity are regulated by ubiquitination. MG132 treatment did not change the levels of ubiquitinated Gsk3␤ (Fig. 1D), suggesting that ubiquitination does not trigger Gsk3␤ degradation. Consistent with it, the cycloheximide-chase assay results showed that WT HA-Gsk3␤ protein and HA-Gsk3␤-KKKK15/27/32/ 36RRRR mutant protein exerted similar patterns of decay following addition and continued incubation with cycloheximide in control-CM or Wnt3a-CM (Fig. 3A), implying that lack of ubiquitination does not alter Gsk3␤ half-life. Next we tested whether ubiquitination regulates Gsk3␤ enzymatic activity. To ensure that the result seen with Gsk3␤ KKKK15/27/32/ 36RRRR mutant was caused by the loss of ubiquitination but not by mutation-caused structural change of the protein, we generated ubiquitin-Gsk3␤ fusion protein that mimics monoubiquitinated Gsk3␤ by fusing one copy of the ubiquitin sequence to the N terminus of HA-Gsk3␤-KKKK15/27/32/ 36RRRR mutant (Fig. 3B). The ubiquitin fusion protein approach, complementary to the ubiquitination mutant approach, has been widely used to validate the functional significance of substrate ubiquitination, for example, the role of monoubiquitination in control of p53 fate (24). shRNA knockdown of ␤-TrCP did not affect ubiquitin fused to Gsk3␤-KKKK15/27/32/36RRRR (Fig. 3C). Using the fusion protein as a tool, we performed an in vitro kinase assay. The results showed that mutant Gsk3␤ proteins, both KKKK15/27/32/36RRRR and ub-KKKK15/27/32/36RRRR, phosphorylated ␤-catenin in vitro as efficiently as WT Gsk3␤ (Fig. 3D). Gsk3␤ phosphorylates LRP6 at Ser-1490 (10). Consistent with the in vitro result, we found that overexpression of WT Gsk3␤ and the mutants exerted similar effects on phosphorylation of LRP6 at Ser-1490 in HEK293T cells (Fig. 3E). Together, these results indicated that the mutant proteins have proper native folding or conformation and that Gsk3␤ enzymatic activity is not altered by its ubiquitination status.
The co-immunoprecipitation results showed that Gsk3␤-␤-TrCP binding was continuously increased by Wnt stimulation (Fig. 4C). Wnt-induction of Gsk3␤-␤-TrCP interaction was further confirmed via the results of the reciprocal IP and IB (Fig.  4D). It is generally believed that ␤-TrCP recognizes its targets in a phosphorylation-dependent manner through a conventional recognition site DpSGxxpS or through an unconventional recognition site (26,27). We speculated that Wnt induces Gsk3␤-␤-TrCP interaction through two mechanisms: (i) Wnt induces Gsk3␤ phosphorylation at particular site(s), which triggers the recruitment of ␤-TrCP to Gsk3␤ and (ii) ubiquitination of Gsk3␤ in turn further increases ␤-TrCP binding to Gsk3␤. Identification of the putative phosphorylation sites is beyond the scope of this study; therefore we have focused on the second mechanism using Gsk3␤ mutants as an investigating tool.
We next tested if Gsk3␤ directly binds to ␤-TrCP by mixing purified human recombinant ␤-TrCP with in vitro synthesized wt Gsk3␤ or Gsk3␤ mutant proteins. The in vitro binding results (Fig. 4I) mirrored the in vivo binding result (Fig. 4F), confirming a direct binding between Gsk3␤ and ␤-TrCP. The in vitro binding result also raised a question why the affinity of in vitro synthesized wt HA-Gsk3␤ for GST-TrCP was substantially higher than that of HA-Gsk3␤-KKKK15/27/32/36RRRR mutant protein? It is well documented that the in vitro generated proteins using rabbit reticulocyte lysates are post-translationally modified similarly as cellular proteins (28), so a potential explanation for this is that the in vitro translated Gsk3␤ proteins might be modified similarly as cellular Gsk3␤. Consistent with this notion, Western blot analysis showed that HA-Ub-Gsk3␤ KKKK15/27/32/36RRRR had two bands (Fig.  4I). As indicated by the results in Fig. 3B, the upper band represented ubiquitinated HA-Ub-Gsk3␤ KKKK15/27/32/ 36RRRR. This proof-of-principle data demonstrated that in vitro translated Gsk3␤ can be ubiquitinated.
The dominant paradigm of ␤-TrCP-mediated ubiquitination is degradation of the ubiquitinated substrate (22). To the best of our knowledge, this is the first evidence that ␤-TrCP-mediated ubiquitination does not lead to substrate degradation. Strikingly, induction of substrate ubiquitination and substrate-␤-TrCP binding is used by the cells as a novel mechanism to inhibit ubiquitination and degradation of another substrate within the same complex.
Ubiquitination of Gsk3␤/Gsk3␤-␤-TrCP Interaction Is Required for Late Wnt Signaling-We next validated the role of Gsk3␤ ubiquitination/Gsk3␤-␤-TrCP binding in Wnt/␤catenin signaling. We first conducted course experiments (up to 16 h after Wnt3a simulation) in which levels of ubiquitinated Gsk3␤, phosphorylated ␤-catenin, and overall levels of ␤-catenin were determined in parallel in wt MEFs. Consistent with the result showing in Fig. 1B, we found that Wnt3a treatment induced Gsk3␤ ubiquitination in MEFs (Fig. 5A). Induction of monoubiquitination of Gsk3␤ did not correlate with Wnt-induced early phase of ␤-catenin accumulation; however, it coincided with Wnt-induced late phase (about 2 h after Wnt3a treatment) stabilization of ␤-catenin (Fig. 5A). This result indicated that Gsk3␤ ubiquitination is involved in chronic (but not acute) Wnt signaling.
Mutations of APC gene occur in 85% of human colorectal cancer (34). The current study shows that ubiquitination of Gsk3␤ regulates Wnt/␤-catenin signaling in HEK293T cells and MEFs (with intact APC protein) and SW480 cells (bearing truncating APC mutations), implying that regulation of Wnt signaling by Gsk3␤ ubiquitination is independent of APC status. Our data suggest that ubiquitination of Gsk3␤ provides a platform for non-Wnt pathway components, directly or indirectly, to be involved in the regulation of Wnt signaling despite the loss of functional APC. Our finding that Wnt signaling requires Gsk3␤ ubiquitination/Gsk3␤-␤-TrCP interaction could have far-reaching implications for identifying novel approaches for targeting Wnt signaling regardless of the status of APC protein.