Synergistic Activation of the Wnt Signaling Pathway by Dvl and Casein Kinase I (cid:1) *

Although casein kinase I (cid:1) (CKI (cid:1) ) has been shown to regulate the Wnt signaling pathway positively, its mode of action is not clear. In this study we show that CKI (cid:1) activates the Wnt signaling pathway in co-operation with Dvl. CKI (cid:1) and Axin associated with different sites of Dvl, and CKI (cid:1) and Dvl interacted with distinct regions on Axin. Therefore, these three proteins formed a ternary complex. Either low expression of Dvl or CKI (cid:1) alone did not accumulate (cid:2) -catenin, but their co-expres-sion accumulated greatly. Dvl and CKI (cid:1) activated the transcriptional activity of T cell factor (Tcf) synergistically. Although the Dvl mutant that binds to Axin but not to CKI (cid:1) activated Tcf, it did not synergize with CKI (cid:1) . Another Dvl mutant that does not bind to Axin did not activate Tcf irrespective of the presence of CKI (cid:1) . Fur-thermore, Dvl and CKI (cid:1) co-operatively induced axis duplication of Xenopus embryos. These results indicate that Dvl and CKI (cid:1) synergistically activated the Wnt signaling pathway and that the binding of the complex of Dvl and CKI (cid:1) to Axin is necessary for their synergistic action. action. These results demonstrate that the binding of the complex of CKI (cid:3) and Dvl to Axin activates the Wnt signaling pathway. construct by chain reaction with Xba and Sma I was digested with Xba and Sma and into the Xba I- and Sma I-cut pCGN. construct pSP64T-Myc/hCKI (cid:3) was digested with Hin dIII and blunted with Klenow fragment. 1.3-kb fragment pSP64T was digested with Spe blunted Klenow. pBJ- Myc/rAxin

Wnt proteins constitute a large family of cysteine-rich secreted ligands that control development in organisms ranging from nematode worms to mammals (1). In vertebrates, the Wnt signaling pathway regulates axis formation, organ development, and cellular proliferation, morphology, motility, and fate (2)(3)(4)(5). In unstimulated cells, free cytoplasmic ␤-catenin is destabilized by a multiprotein complex containing Axin (or its homolog Axil/conductin), GSK-3␤, 1 and APC (6 -13). Axin functions as a scaffold protein in this complex by directly binding to GSK-3␤, ␤-catenin, and APC. Interaction of GSK-3␤ with Axin in the complex facilitates efficient phosphorylation of ␤-catenin by GSK-3␤. Phosphorylated ␤-catenin forms a complex with Fbw1 (␤TrCP/FWD1), a member of the F-box protein family, resulting in the degradation of ␤-catenin by the ubiquitin and proteasome pathways (14,15). Because Axin inhibits Wnt-dependent accumulation of ␤-catenin and activation of Tcf/Lef, a transcription factor (10,16), it is a negative regulator of the Wnt signaling pathway. In addition, APC and Axin are also phosphorylated by GSK-3␤ in the Axin complex. Phosphorylation of APC enhances its binding to ␤-catenin (17), and that of Axin stabilizes it, in contrast to phosphorylation of ␤-catenin (18).
When cells are stimulated by Wnt, a cytoplasmic protein, Dvl, antagonizes the action of GSK-3␤. Although whether Dvl binds directly to Frizzled, the receptor for Wnt, or whether intermediary proteins are involved in the signal transmission between Frizzled and Dvl is not known at present, Dvl appears to bind to the Axin complex (19 -21) and to inhibit GSK-3␤-dependent phosphorylation of ␤-catenin, APC, and Axin (18,20,22). Once the phosphorylation of ␤-catenin is reduced, it dissociates from the Axin complex, and ␤-catenin is no longer degraded, resulting in its accumulation in the cytoplasm. Stabilized ␤-catenin is translocated into the nucleus, where it binds to Tcf/Lef (23)(24)(25) and serves as a coactivator of Tcf/Lef to stimulate transcription of the Wnt target genes including cmyc, fra, jun, cyclin D1, peroxisome proliferator-activated receptor ␦, and matrilysin (26 -31). Thus, the Wnt signal stabilizes ␤-catenin, thereby regulating various gene expression.
Three Dvl genes, Dvl-1, -2, and -3, have been isolated in mammals (32)(33)(34). Expression of Dvl in cells induces the accumulation of ␤-catenin and the activation of Tcf (20,21,35). Dvl homologs are conserved in Drosophila (dishevelled, Dsh) and Xenopus (Xenopus dishevelled, Xdsh) (36 -38). All Dsh and Dvl family members contain three highly conserved domains (2)(3)(4): an N-terminal DIX domain, which is also found in the C terminus of Axin; a central PDZ domain, which has been shown to be a protein-protein interaction surface in several proteins; and a DEP domain, which is conserved in proteins that regulate GTP-binding proteins. The DIX domain is necessary for the Dvl activity to regulate the Wg and Wnt signaling pathways positively (20, 39 -41). Disruption of the PDZ domains of Dsh and Dvl abolishes their activities in the Wg and Wnt signaling pathways and in the Xenopus secondary axis formation, suggesting that the PDZ domain is essential for the Wnt signaling pathway (20,38,42). Dvl antagonizes the ability of Axin to induce ventralization in Xenopus embryos (43), and the DIX and PDZ domains of Dvl are important for its complex formation with Axin (19 -21). The DEP domain of Dsh has been found to be critical for rescue of the Drosophila Dsh planar cell polarity defect and for the activation of c-Jun N-terminal kinase but not essential for the Wg pathway (44,45). The DEP domain of vertebrate Dvl is also necessary for c-Jun N-terminal kinase activation in mammals but not for the axis formation (39,46). CKI comprises a large family of related gene products, ␣, ␤, ␥, ␦, and ⑀ (47). Each shares at least 50% amino acid identity within the protein kinase catalytic domain. Distinct CKI family members are likely to show different tissue distributions and subcellular localization and to have distinct roles including the regulation of DNA repair, DNA replication, cell cycle progression, and circadian rhythm (47,48). It has been reported that among CKI family, CKI⑀ and ␦ but not CKI␣ are involved in the Wnt signaling pathway (49,50). Overexpression of CKI⑀ in Xenopus embryos induces the expression of siamois, a Wnt response gene, and axis duplication. These CKI⑀-dependent responses are suppressed by Axin and GSK-3␤, and Dvl-induced axis duplication is inhibited by CKI-7, a CKI inhibitor (49). Furthermore, CKI⑀ forms a complex with Dvl and Axin, and CKI⑀ activates Lef-1, which is inhibited by Axin in mammalian cells (50). These results suggest that CKI⑀ positively regulates the Wnt signaling pathway by functioning between Dvl and GSK-3␤. However, the mode of action of CKI⑀ is not understood.
Here we demonstrate that by complex formation with Axin, CKI⑀ and Dvl synergistically accumulate ␤-catenin and activate Tcf in mammalian cells and that they also synergistically induce axis duplication in Xenopus embryos. Furthermore, we show that the binding of CKI⑀ to Dvl is necessary for their synergistic action. These results demonstrate that the binding of the complex of CKI⑀ and Dvl to Axin activates the Wnt signaling pathway.
Immunocytochemistry-L cells grown on coverslips were fixed for 10 min in PBS containing 4% paraformaldehyde. The cells were washed with PBS three times and then permeabilized with PBS containing 0.1% Triton X-100 and 2 mg/ml bovine serum albumin for 2 h. The cells were washed and incubated for 1 h with the anti-HA and anti-␤-catenin antibodies. After being washed with PBS, they were further incubated for 1 h with Cy5-labeled anti-mouse IgG and Alexa 546-labeled antirabbit IgG. The coverslips were washed with PBS, mounted on glass slides, and viewed with a confocal laser-scanning microscope (LSM510, Carl-Zeiss, Jena, Germany).

Xenopus Injections and Analyses of Phenotypes-Myc-tagged hCKI⑀
and Dvl-1 cDNA were subcloned into pSP64T (57). Sense mRNA was obtained by in vitro transcription of linearized templates using the SP6-mMESSAGE mMACHINE kit (Ambion, Austin, TX). Fertilized eggs were de-jelled using 4.5% L-cysteine hydrochloride monohydrate, and mRNAs were injected into ventral blastomeres at the four-cell stage. After injection, embryos were cultured for 3 days (stage 40 -41).
Others-Protein concentrations were determined with bovine serum albumin as a standard (58).

Complex Formation of CKI⑀ with Dvl or Axin at Endogenous
Level-Deletion mutants of rAxin and Dvl-1 used in this study are shown in Fig. 1. CKI⑀ was previously shown to form a complex with Dvl and Axin (49,50). However, because these experiments were done with overexpression assays in mammalian cells and a yeast two-hybrid system, we first examined whether CKI⑀ associates with Dvl and Axin at endogenous level in intact cells. To this end, we generated the antibodies that immunoprecipitate Dvl and Axin. CKI⑀ was detected slightly but reproducibly in the Dvl immune complex from L cells (Fig. 2, lanes 1-3). In the Axin complex immunoprecipitated from L cells, CKI⑀ was observed in addition to GSK-3␤ (Fig. 2, lanes 4 -6). Because endogenous Dvl was not detected in this Axin complex (data not shown), it is likely that the complex formation of CKI⑀ with Axin is not mediated via Dvl in the condition without Wnt stimulation. These results indicate that CKI⑀ at least forms a complex with Dvl or Axin at endogenous level in intact cells.
Complex Formation of Dvl with CKI⑀ -HA-Dvl-1 was coexpressed with GFP-CKI⑀ and GFP-CKI⑀ (KN) in COS cells. When the lysates were immunoprecipitated with the anti-HA antibody, GFP-CKI⑀ and GFP-CKI⑀ (KN) were co-precipitated with HA-Dvl-1 with a similar efficiency (Fig. 4A), suggesting that kinase activity of CKI⑀ is not required for its complex formation with Dvl. To examine which region of Dvl is responsible for the complex formation with CKI⑀, various deletion mutants of HA-Dvl-1 were expressed with GFP-CKI⑀ in COS cells (Fig. 4B, lanes 1-8). When the lysates expressing GFP-CKI⑀ alone were immunoprecipitated with the anti-HA antibody, GFP-CKI⑀ was not observed in the immunoprecipitates (Fig. 4B, lane 9). The PDZ domain of Dvl-1 was shown to be  important for its interaction with CKI⑀ by the yeast two-hybrid assay (49). However, GFP-CKI⑀ was immunoprecipitated with HA-Dvl-1 ⌬PDZ as well as HA-Dvl-1 but not with HA-Dvl-1-(201-371), that contains the PDZ domain (Fig. 4B, lanes 10 -12). The reasons for this discrepancy are not known at present, but it might be due to the difference of assays. Among other deletion mutants, HA-Dvl-1-(1-519) and HA- Dvl-1-(140 -670) but not HA-Dvl-1-(1-378) or HA-Dvl-1-(1-432) formed a complex with GFP-CKI⑀ (Fig. 4B, lanes 13-16). These results suggest that the N-terminal region of Dvl-1 including the DIX and PDZ domains is not important but that the entire DEP domain is necessary for its complex formation with CKI⑀. Taken together with the observations that the N-terminal region of Dvl-1 including the DIX and PDZ domains binds to Axin (20,54,60), these results suggest that the binding sites of Dvl-1 for Axin and CKI⑀ are distinct. To determine that the interaction of Dvl with CKI⑀ is direct, GST-Dvl-1 was incubated with MBP-CKI⑀. GST-Dvl-1 bound to MBP-CKI⑀ (Fig. 4C). In this experiment, equal amounts (20 pmol) of GST-Dvl-1 and MBP-CKI⑀ were used. Although this is not a saturated condition, it seems that ϳ10 -20% of GST-Dvl-1 bound to MBP-CKI⑀, an estimation from the density of a band in the Western blotting. When the kinase activity of CKI⑀ was measured with casein as a substrate, co-expression with HA-Dvl-1 did not affect the CKI⑀ activity (data not shown), suggesting that the interaction with Dvl-1 does not regulate the kinase activity of CKI⑀.
Ternary Complex Formation of Axin, Dvl, and CKI⑀ -The results above demonstrated that Axin and CKI⑀ bind to the distinct regions of Dvl-1 and that Dvl-1 and CKI⑀ form a complex with the different sites of Axin, suggesting that these three molecules form a ternary complex. To clarify this possibility, various combinations of rAxin, Dvl-1, and CKI⑀ were expressed in COS cells. When the lysates expressing Myc-rAxin alone were immunoprecipitated with the anti-Myc antibody, GSK-3␤ was detected in the Myc-rAxin immune complex (Fig.  6, lane 1). Co-expression with either HA-Dvl-1, GFP-CKI⑀, or GFP-CKI⑀ (KN) did not affect the complex formation of Myc-rAxin with GSK-3␤ (Fig. 6, lanes 2, 5, and 6). When Myc-rAxin was co-expressed with both HA-Dvl-1 and GFP-CKI⑀, these two proteins did not compete with each other for their binding to Myc-rAxin (Fig. 6, lane 3). Furthermore, the interaction of Myc-rAxin with both HA-Dvl-1 and GFP-CKI⑀ did not influence its complex formation with GSK-3␤. Co-expression with HA-Dvl-1 and GFP-CKI⑀ (KN) showed the same results (Fig. 6,  lane 4).

Effects of Phosphorylation of APC by CKI⑀ on Its Complex Formation with Axin-As shown in
␤-Catenin was observed in the Myc-APC-(1211-2075) immune complex (Fig. 7, lane 1). Expression of FLAG-rAxin de-  1 and 2). The same lysates (200 g of protein) were immunoprecipitated with the anti-Myc antibody, and the immunoprecipitates were probed with the anti-Myc and anti-CKI⑀ antibodies (lanes 3 and 4). IP, immunoprecipitation; Ab, antibody; WT, wild type. B, interaction of the deletion mutants of rAxin with CKI⑀ in COS cells. The lysates (20 g of protein) of COS cells expressing the indicated proteins were probed with the anti-Myc and anti-CKI⑀ antibodies (lanes 1-9). The same lysates (200 g of protein) were immunoprecipitated with the anti-Myc antibody, and the immunoprecipitates were probed with the anti-Myc and anti-CKI⑀ antibodies (lanes 10 -18). The results shown are representative of four independent experiments. creased the complex formation of ␤-catenin with Myc-APC-(1211-2075) slightly (Fig. 7, lanes 1 and 2). Although this would reflect the down-regulation of ␤-catenin by Axin, a further decrease in the amount of ␤-catenin in the APC-Axin complex by Axin may be small because ␤-catenin complexed with APC is degraded efficiently. In contrast, expression of GFP-CKI⑀ slightly increased the level of ␤-catenin complexed with Myc-APC-(1211-2075), but that of GFP-CKI⑀ (KN) did not (Fig. 7, lanes 3 and 4). Further expression of HA-Dvl-1 increased the amount of ␤-catenin associated with Myc-APC-(1211-2075) (Fig. 7, lane 5). These results suggest that Dvl-1 and CKI⑀ may co-operate to increase ␤-catenin.
Synergistic Effects of Dvl and CKI⑀ on Wnt Signaling-Previously we showed that transient overexpression of Dvl-1 in L cells induces the nuclear accumulation of ␤-catenin (20). However, accumulation of ␤-catenin was not observed in L cells stably expressing Dvl-1 (L/Dvl cells) (Fig. 8, B and D). This result suggests that a low expression level of Dvl-1 in L/Dvl cells is not sufficient for the stabilization of ␤-catenin. Because CKI⑀ may enhance Dvl-1 stimulation of the Wnt signaling pathway (observations in Fig. 7), we examined the effects of combination of CKI⑀ and Dvl-1 on the accumulation of ␤-catenin. Although expression of CKI⑀ in wild-type L cells did not accumulate ␤-catenin (Fig. 8, A and B), its expression in L/Dvl cells increased the level of ␤-catenin in the nucleus (Fig. 8,  C-E). Expression of CKI⑀ (KN) in L/Dvl cells did not induce the accumulation of ␤-catenin (Fig. 8, F-J). These results show that Dvl-1 and CKI⑀ co-operate to accumulate ␤-catenin and that the kinase activity of CKI⑀ is necessary for this synergistic effect.
Synergistic Action of CKI⑀ and Dvl on Axis Duplication-To confirm the mode of action of CKI⑀ and Dvl, we examined their effects on the Wnt signaling pathway using Xenopus embryos. The Wnt signaling pathway regulates axis formation of Xenopus embryos (62). Ventral injection of mRNAs of Xwnt-8, Dvl, and X␤-catenin has been shown to induce the formation of a secondary dorsal axis (38,(62)(63)(64). Consistent with previous observations (49,50), ventral injection of a high dose (200 pg) of CKI⑀ mRNA into embryos resulted in dorsalization of phenotypes such as axis duplication (Fig. 10B). Embryos injected ventrally with either low doses (25 or 50 pg) of Dvl-1 or CKI⑀ mRNA did not exhibit significant abnormalities (Fig. 10, A and  B). However, co-injection of low doses of Dvl-1 and CKI⑀ mRNA induced axis duplication greatly (Fig. 10, A and B). These results demonstrate that CKI⑀ synergistically functions with Dvl-1 to regulate axis formation and are consistent with the results observed in mammalian cells. DISCUSSION In this study we found that CKI⑀ and Dvl form a complex with the different regions of Axin and that CKI⑀ and Axin associate with distinct sites of Dvl. Although the binding of Dvl to CKI⑀ was direct, that of Axin to CKI⑀ was not. Furthermore, our results demonstrated that the simultaneous binding of Dvl and CKI⑀ to Axin induces the maximal activation of the Wnt signaling pathway and that the direct binding of CKI⑀ to Dvl is necessary for this synergistic activation.
We previously showed that the N-terminal region of Dvl-1 including the DIX domain and the N-terminal half of the PDZ domain is important for its binding to Axin and that the PDZ domain is essential for the Wnt signaling pathway (20). For instance, disruption of the region containing the PDZ domain (Dvl-1-(282-336)) reduces its ability to stabilize ␤-catenin (20). This is consistent with the previous observations that the PDZ domain is important for Dvl to induce Xenopus axis duplication (38) and to regulate Drosophila segment polarity (41). Therefore, it has been speculated that there is a protein that regulates the Wnt signaling pathway by binding to the PDZ domain of Dvl. Although CKI⑀ has been reported to bind to the PDZ domain of Dvl-1 by a yeast two-hybrid assay, our results using intact mammalian cells showed that the PDZ domain is not necessary for the binding of Dvl-1 to CKI⑀. We do not know the exact reasons, but it might be due to the differences between yeast two-hybrid and mammalian intact cell assays.
Our results showed that CKI⑀ and Dvl-1 co-operate to stimulate ␤-catenin accumulation and activate Tcf in mammalian cells and to induce axis duplication in Xenopus embryos. These synergistic activities between CKI⑀ and Dvl-1 were observed when Dvl-1-(1-519) but not Dvl-1-(1-432) was used. Since Dvl-1-(1-519) but not Dvl-1-(1-432) forms a complex with CKI⑀, these results suggest that the entire DEP domain of Dvl-1 is necessary for its binding to and synergistic action with CKI⑀. Dvl-1-(140 -670), which binds to CKI⑀ but weakly to Axin, did not activate Tcf, suggesting that the binding of CKI⑀ to Dvl-1 is not sufficient for the activation of the Wnt pathway but that the binding of the DIX domain of Dvl-1 to Axin is necessary. Furthermore, Dvl-1 ⌬PDZ , which binds to both CKI⑀ and Axin, did not activate Tcf, suggesting that the binding of Dvl to Axin is not sufficient for the activation of the Wnt signaling pathway. Therefore, the protein that binds to the PDZ domain of Dvl-1 and activates the Wnt signaling pathway remains to be clarified. Taken together, these results indicate that the binding of the complex including Dvl-1, CKI⑀, and at least one more protein to Axin is important for the activation of the Wnt signaling pathway.
Although the mechanism by which Dvl activates the Wnt signaling pathway is not understood, it has been suggested that Frat, which was identified as a GSK-3␤-binding protein (65), forms a complex with the PDZ domain of Dvl and that this complex induces the dissociation of GSK-3␤ from Axin in response to Wnt (66). Furthermore, it has been shown that expression of Dvl induces the displacement of GSK-3␤ from the ARP (Axin-related protein) and that the PDZ domain is necessary for this Dvl activity (61). In addition to Frat, CKII (67) and protein phosphatase 2C (68) have been reported to form a complex with the PDZ domain of Dvl. Since none of these proteins contains the (S/T)XV sequence in their C termini (69), the PDZ domain of Dvl may have different properties from that of other known PDZ domains regarding protein-protein interactions. CKII associates with and phosphorylates Dvl (67). Although the interaction with CKII requires the region containing the PDZ domain, whether this interaction is direct is not known, and its significance in the Wnt signaling pathway is not yet clear. Protein phosphatase 2C has been isolated by the yeast two-hybrid assay using the PDZ domain as bait (68). Expression of protein phosphatase 2C in COS cells dephosphorylates and down-regulates Axin and stimulates the transcriptional activity of Lef-1. These results suggest that protein phosphatase 2C works as a positive regulator of the Wnt signaling pathway by inhibiting phosphorylation in the Axin complex. We have recently identified a novel Dvl-binding protein, Idax, that suppresses the Wnt signaling (54). Idax interacts with the PDZ domain of Dvl and inhibits the complex formation of Dvl and Axin. It inhibits Wnt-dependent accumulation of ␤-catenin in mammalian cells and Wnt-dependent axis duplication in Xenopus embryos. Therefore, Idax negatively regulates the Wnt signaling pathway probably by inhibiting the binding of Axin to Dvl or by competing with other PDZ domain-binding proteins for their interaction with Dvl. It is necessary to examine whether CKI⑀ affects the complex formation of Frat, CKII, protein phosphatase 2C, or Idax with Dvl and regulates their functions.
How does CKI⑀ activate the Wnt signaling pathway? To accumulate ␤-catenin in the cells, the phosphorylation of ␤-catenin must be reduced. The simple model is that CKI⑀ induces the dissociation of GSK-3␤ from Axin by phosphorylating substrates in the Axin complex or that CKI⑀ inhibits the GSK-3␤ activity. Dvl-1, APC, Axin, and ␤-catenin were good substrates of CKI⑀ in vitro, and at least Dvl-1 and APC were phosphorylated by CKI⑀ in intact cells. However, CKI⑀ did not affect the interaction of Dvl or APC with Axin. The CKI⑀ complex contained GSK-3␤ via Axin, and CKI⑀ did not dissociate GSK-3␤ from Axin. We also examined whether CKI⑀ regulates GSK-3␤ activity. CKI⑀ did not phosphorylate GSK-3␤ directly. The activity of GSK-3␤ complexed with CKI⑀ via Axin was the same as that complexed with CKI⑀ (KN) (data not shown). Dvl did not affect the activity of GSK-3␤ complexed with CKI⑀, either (data not shown). Therefore, it is unlikely that CKI⑀ regulates the localization and activity of GSK-3␤ in the Axin complex. Another possibility is that an unknown protein might be phosphorylated by CKI⑀ and that it may prevent the ubiquitination of ␤-catenin.
The complex formations of CKI⑀ and Dvl or CKI⑀ and Axin at endogenous levels were small. Furthermore, we could not detect the endogenous association of Dvl with Axin. Therefore, it is likely that the complexes of Axin, Dvl, and CKI⑀ are less stable than that of Axin and GSK-3␤. This is reasonable, because ␤-catenin is usually degraded in the Axin-GSK-3␤ complex and accumulated in response to Wnt. The more stable complex of Axin, Dvl, and CKI⑀ may be formed once Wnt stimulates the cells. It is possible that the interactions we observed with overexpressed proteins are representative of those in the cells activated by Wnt. We are examining the possibility that Wnt stabilizes the complex of Axin, Dvl, and CKI⑀, thereby accumulating ␤-catenin.
In summary, CKI⑀ and Dvl synergistically activate the Wnt signaling pathway by their complex formation with Axin, and the binding of CKI⑀ to Dvl is important for their synergistic effects. In addition to them, one more protein that binds to the PDZ domain of Dvl-1 would be essential for the activation of the Wnt signaling pathway. CKI⑀ could be located to the proper position via Axin and Dvl and phosphorylate substrates in the Axin complex, resulting in the accumulation of ␤-catenin. Further studies will be necessary for understanding the whole picture of the mechanism by which Dvl and CKI⑀ regulate the Wnt signaling pathway positively.