Akt participation in the Wnt signaling pathway through Dishevelled.

Inactivation of glycogen synthase kinase 3beta (GSK3beta) and the resulting stabilization of free beta-catenin are critical steps in the activation of Wnt target genes. While Akt regulates GSK3alpha/beta in the phosphatidylinositide 3-OH kinase signaling pathway, its role in Wnt signaling is unknown. Here we report that expression of Wnt or Dishevelled (Dvl) increased Akt activity. Activated Akt bound to the Axin-GSK3beta complex in the presence of Dvl, phosphorylated GSK3beta and increased free beta-catenin levels. Furthermore, in Wnt-overexpressing PC12 cells, dominant-negative Akt decreased free beta-catenin and derepressed nerve growth factor-induced differentiation. Therefore, Akt acts in association with Dvl as an important regulator of the Wnt signaling pathway.

Inactivation of glycogen synthase kinase 3␤ (GSK3␤) and the resulting stabilization of free ␤-catenin are critical steps in the activation of Wnt target genes. While Akt regulates GSK3␣/␤ in the phosphatidylinositide 3-OH kinase signaling pathway, its role in Wnt signaling is unknown. Here we report that expression of Wnt or Dishevelled (Dvl) increased Akt activity. Activated Akt bound to the Axin-GSK3␤ complex in the presence of Dvl, phosphorylated GSK3␤ and increased free ␤-catenin levels. Furthermore, in Wnt-overexpressing PC12 cells, dominant-negative Akt decreased free ␤-catenin and derepressed nerve growth factor-induced differentiation. Therefore, Akt acts in association with Dvl as an important regulator of the Wnt signaling pathway.
Wnt signaling regulates several developmental processes in both insects and vertebrates. Aberrant Wnt signaling occurs both in oncogenic processes (1)(2)(3) and in some cardiovascular diseases (4 -6). The interaction between secreted Wnt glycoproteins and their cognate Frizzled receptors leads to the activation of Dvl 1 protein, which transmits the Wnt signal to downstream effectors. By inhibiting phosphorylation of ␤-catenin by GSK3␤, activated Dvl prevents the ubiquitination and subsequent proteosomal degradation of ␤-catenin. The resulting accumulation of free ␤-catenin enhances its interaction with transcription factors of the lymphoid enhancer factor-T cell factor (LEF/TCF) family and induces the transcription of target genes such as cyclin D1 and c-myc (1)(2)(3). Two mechanisms have been advanced to account for the inhibition of GSK3␤ by Dvl: 1) the catalytic activity of GSK3␤ is blocked by serine phosphorylation (1,(7)(8)(9)(10); 2) the interaction of GSK3␤ and ␤-catenin is disrupted by a conformation change of the Axin-␤-catenin-GSK3␤-adenomatosis polyposis coli protein complex (Axin complex) (2,(11)(12)(13). However, in the first instance, the process whereby Dvl stimulates serine phosphorylation has not been elucidated; nor is it clear whether these two mechanisms function in a concerted manner to promote Wnt signaling.
The serine/threonine kinase Akt (protein Kinase B/related to A and protein kinase C) is a major effector of the PI3K pathway and is activated by many polypeptide growth factors (14 -16). The recruitment of Akt from the cytoplasm to the plasma membrane by the lipid products of PI3K leads to Akt phosphorylation by 3-phosphoinositide-dependent protein kinases. This phosphorylation of Akt at Thr 308 and Ser 473 results in its activation. Akt acts in part through its phosphorylation of GSK3␣/␤, which in turn regulates cell metabolism. Recognizing the pivotal role of GSK3␤ in Wnt signaling, we tested the hypothesis that Akt functions in this pathway to control GSK3␤ activity.
Wnt-3a-conditioned Medium-Wnt-3a-and control-conditioned medium were prepared essentially as described (21). We verified that Wnt-3a medium specifically up-regulated the cellular ␤-catenin levels in L cells.
Reporter Assays-293T cells were transfected using Lipo-fectAMINE (Life Technologies, Inc.) in 6-well plates with 30 ng/well of TOPFlash or FOPFlash luciferase reporter (Upstate Biotechnology) together with effector plasmids or empty pcDNA3 vector (amount added to keep the total DNA content at 1 g/well) as described in the figure legends. Cell extracts were prepared 24 h after transfection by a deter-  1 The abbreviations used are: Dvl, Dishevelled; GSK3␤; glycogen synthase kinase 3␤; LEF, lymphoid enhancer factor; TCF, T cell factor; PI3K, phosphatidylinositide 3-OH kinase; DMEM, Dulbecco's modified Eagle's medium; FBS, fetal bovine serum; PCR, polymerase chain reaction; GST, glutathione S-transferase; EGFP, enhanced green fluorescent protein; NGF, nerve growth factor; sFRP-1, Frizzled-related protein-1. gent lysis method (Promega). Transfection efficiencies were normalized by pCMV␤-gal reporter activity (CLONTECH). Each construct was transfected at least three times in triplicate. Reporter activities (mean Ϯ S.D.) are presented as the -fold increase of TOPFlash activity from the cells transfected with empty vector.
Immunoprecipitation and Akt Kinase Activity Assays-Cell lysates were precleared with protein A or G-Sepharose beads (Amersham Pharmacia Biotech or Calbiochem, respectively) for 1 h at 4°C, and then incubated with the indicated antibody and protein A or G-Sepharose beads for 3.5 h at 4°C. The immunocomplexes were pelleted and washed three times with cold lysis buffer. The samples were subjected to immunoblotting with the indicated antibodies. Akt activity was measured using an Akt kinase assay kit (New England BioLabs) with recombinant GSK3 as a substrate for immunoprecipitated Akt.
Analysis of Free ␤-Catenin Levels-Free ␤-catenin levels were detected by GST-E-cadherin pull-down assay as described (18). GST-Ecadherin fusion protein was expressed in bacteria and purified by binding to glutathione-Sepharose beads (Amersham Pharmacia Biotech). 1 mg of cell lysate was incubated with GST-E-cadherin-beads for 1 h at 4°C. The beads were collected by centrifugation and washed, and proteins were dissolved in Laemmli buffer and subjected to SDS-polyacrylamide gel electrophoresis followed by immunoblotting with anti-␤-catenin antibody.
Neuronal Differentiation-Cells were incubated in differentiation medium (DMEM supplemented with 1% horse serum and 100 ng/ml NGF (Austral Biologicals)) or growth medium (DMEM supplemented with 10% horse serum, 5% FBS). Medium was refreshed every 48 h. Neurite out-growth was defined as a phase dark process with a clearly defined growth cone that was at least 1.5 cell diameters in length. Experiments were carried out in triplicate and repeated at least three times, and in each experiment cells were counted by at least two individuals who were blinded to the groups.
Statistics-Where indicated, comparisons between groups were made by factorial analysis of variance followed by Scheffe's test (

Wnt Signaling Induces Phosphorylation and Activation of
Akt-We first explored the possibility that Wnt signaling stimulates Akt activity. In a Wnt-1-overexpressing PC12 cell line (Int5) (17), but not in either the parental PC12 or in an antisense Wnt-1-expressing PC12 cell line (Tni3), Akt was phosphorylated at both Thr 308 and Ser 473 (Fig. 1A). The phosphorylation of both residues is necessary and sufficient for full Akt activation (16). Accordingly, Akt activity, measured by its ability to phosphorylate recombinant GSK3␤, was much higher in Int5 cells than in control cells (Fig. 1A). Furthermore, phosphorylation of GSK3␤ at Ser 9 , which results in its inactivation (10), was higher in Int5 than in control cells (Fig. 1, A and B). Phosphorylation of Akt and GSK3␤ was inhibited by the Wnt antagonist-secreted Frizzled-related protein-1 (sFRP-1) (Fig.  1C) accompanied by a reduction in free ␤-catenin as we showed previously (22). These results indicated that phosphorylation of Akt and GSK3␤, and free ␤-catenin levels in Int5 cells, were dependent on secreted Wnt-1 protein. To extend these observations to other cells, we also analyzed Akt phosphorylation in NIH-3T3 cells transiently transfected with Wnt-1, Dvl-1, and a ␤-catenin mutant (S37A ␤-catenin). Previous studies had shown that overexpression of these gene products cause an increase in free ␤-catenin, indicative of Wnt signaling (2,8,(23)(24)(25). Transient expression of Wnt-1 or Dvl-1, but not S37A ␤-catenin, increased Akt phosphorylation (Fig. 1D). These results indicated that Akt activation is a consequence of Wnt signaling that occurs downstream of Dvl.
Dominant-negative Akt Inhibits Wnt Signaling Pathway-To determine whether activated Akt is required for the canonical Wnt signaling, we used a dominant-negative form of Akt (AAA-Akt) (26). Int5 cells infected with an adenovirus expressing AAA-Akt (Ad.AAA-Akt) showed reductions in phosphorylated GSK3␤ and free ␤-catenin protein levels ( Fig. 2A). These changes did not occur in cells infected with control virus (Ad.␤gal.EGFP). In transient transfection assays using 293T cells, Wnt-1, Dvl-1, and S37A ␤-catenin each enhanced TCF-binding site reporter gene activity (TOPFlash) (Fig. 2B). AAA-Akt suppressed TOPFlash activity induced by Wnt-1 or Dvl-1 but did not affect the S37A ␤-catenin induced activity (Fig. 2B). The activity of the mutated TCF-binding site reporter (FOPFlash) was not affected in these experiments (data not shown and Fig.  3B). These results implied that Akt is involved in Wnt signaling, functioning at a point in the pathway between Dvl and ␤-catenin. Akt Up-regulates Free ␤-Catenin Levels and Wnt Target Gene Expression in the Presence of Dvl-To test whether Akt activity is sufficient to stimulate an increase in free ␤-catenin, we infected Int5 and Tni3 cells with either an adenovirus-expressing wild type Akt (Ad.Akt) or a control adenovirus (Ad.EGFP). Although the amount and activity of Akt protein were similarly increased in both lines infected with Ad.Akt, a rise in free ␤-catenin levels was seen only in the Int5 cells (Fig. 3A). These results demonstrated that Akt activity was not sufficient to up-regulate free ␤-catenin but that it could augment the activity of a Wnt stimulus.
Recalling that a dominant-negative form of Akt inhibited Wnt signaling induced by Dvl, we postulated that Akt might act in conjunction with Dvl to promote Wnt signaling. Overexpression of Akt in 293T cells produced a dose-dependent increase in Akt activity (Fig. 3B). As observed with PC12 cell derivatives (Fig. 3A), Akt activity was not sufficient to induce Wnt signaling in the reporter assay. However, Akt enhanced TOPFlash activity that was stimulated by Dvl-1 (Fig. 3B) but not by S37A ␤-catenin (data not shown). Neither Akt nor Dvl-1 affected control FOPFlash activity (Fig. 3B). These results suggested that Dvl is required for Akt to up-regulate Wnt-dependent gene expression.
Akt Interacts with the Axin Complex in the Presence of Dvl-To delineate the mechanism of Akt and Dvl signal trans-duction, we investigated their interaction with the Axin complex. Others have shown that activated Dvl binds Axin and recruits GSK3␤-binding protein to the Axin complex. This binding induces a conformational change in the complex, resulting in the inhibition of GSK3␤ phosphorylation of ␤-catenin (2,12,13). We first investigated the association between endogenous Akt and Axin. Similar to Wnt-1, Wnt-3a-conditioned medium induces the canonical Wnt signaling pathway (2,9,25). Axin was co-immunoprecipitated with Akt in 293T cells treated for 24 h with Wnt-3a-conditioned medium but not control medium (Fig. 4A). Thus, the association of Akt with Axin requires a Wnt stimulus. Consistent with results using Wnt-1overexpressing PC12 cells (Fig. 1A), these Wnt-3a-treated 293T cells had increased phosphorylation levels of Akt and GSK3␤ (Fig. 4A, Cell lysate). We also examined total GSK3␤ in cell lysates and GSK3␤ in the Axin complex, the latter defined as the GSK3␤ protein co-immunoprecipitated with Axin. In transfected 293T cells, Dvl expression stimulated the phosphorylation of both total GSK3␤ (Fig. 4B, Cell lysate) and GSK3␤ in the Axin complex (Fig. 4B, IP). Akt expression alone did not affect the phosphorylation of GSK3␤ in the Axin complex (Fig. 4B,  IP), although it did stimulate the phosphorylation of total GSK3␤ (Fig. 4B, Cell lysate). However, in the presence of Dvl, Immunoprecipitation and immunoblotting were performed as described above. D, Akt levels are correlated with GSK3␤ levels in the Axin complex. 293T cells were transfected with Dvl-Myc and Akt and with GSK3␤ and/or Axin-FLAG as indicated. Immunoprecipitation and immunoblotting were performed as described above.
Akt increased the amount of phosphorylated GSK3␤ in the Axin complex (Fig. 4B, IP). These results suggested that Dvl was necessary for Akt-dependent phosphorylation of GSK3␤ in the Axin complex. Furthermore, phosphorylated Akt was included in the Axin complex only in the presence of Dvl (Fig. 4B).
Dvl protein family members contain three highly conserved domains: an N-terminal DIX (Dishevelled-Axin) domain, a central PDZ (PSD-95-Discs-large-ZO-1) domain, and a C-terminal DEP (Dishevelled-EGL-10-Pleckstrin) domain (13,27). Although the roles of each domain have not been elucidated completely, either the DIX or PDZ domain is sufficient for Axin binding (11). To confirm that Dvl binding to Axin is necessary for Akt involvement in the Axin complex, Dvl mutants were generated. Consistent with previous reports (11,28), C-Dvl-(369 -695), which lacks both the DIX and PDZ domains, did not bind to Axin, whereas N-Dvl-(1-497), which lacks the C-terminal region adjacent to the DEP domain, bound to Axin (Fig.  4C). Full-length Dvl and N-Dvl-(1-497), but not C-Dvl-(369 -695), induced Akt interaction with the Axin complex (Fig. 4C). Although the magnitude of Akt association with the Axin complex induced by N-Dvl (1-497) was less than that seen with full-length Dvl-1, this difference was consistent with the relative amounts of the recombinantly expressed Dvl proteins (Fig.  4C). C-Dvl-(369 -695) and N-Dvl-(1-497) were expressed at similar levels (Fig. 4C), indicating that the absence of Akt in the Axin complex is not due to reduced C-Dvl-(369 -695) expression levels. These results suggest that the N-terminal region of Dvl, which can bind to Axin, is required to induce Akt association with the Axin complex.
Akt did not interact directly with Axin or Dvl ( Fig. 4B and data not shown). However, Akt binding to the Axin complex may be mediated by GSK3␤, judging from the correlation of Akt and GSK3␤ levels in the complex (Fig. 4D). As shown in Fig. 4D, the amount of GSK3␤ and Akt protein in the Axin complex was higher in GSK3␤-overexpressing cells than in control cells. Taken together, our results indicate that Dvl binding to the Axin complex enables Akt to associate with and to phosphorylate GSK3␤ in the complex.
Dominant-negative Akt Restores Neuronal Differentiation of Int5 Cells-Wnt signaling frequently is responsible for changes in cell morphology and differentiation (1)(2)(3). For instance, the Wnt-overexpressing Int5 cells have a flat shape and do not differentiate with NGF treatment, whereas the parental PC12 cells are round and differentiate readily in response to NGF (17). sFRP-1-treated Int5 cells differentiated in response to NGF (Fig.  5A); this implies that the phenotype of Int5 cells depends on Wnt signaling. To test the impact of Akt on this Wnt-dependent cellular phenotype, we infected Int5 cells with either a dominantnegative Akt (Ad.AAA-Akt) or a control (Ad.␤-gal.EGFP) adenovirus. In contrast to the cells treated with control virus, Int5 cells infected with Ad.AAA-Akt had a more differentiated appearance with extensive neurite outgrowth in response to NGF (Fig. 5B and C). Neuronal differentiation of Int5 cells was confirmed by changes in peripherin levels. Consistent with a previous report (29), in response to NGF, Wnt-1-overexpressing PC12 cells (Int5) did not up-regulate peripherin protein expression, whereas parental PC12 cells did (Fig. 5D). However, Ad.AAA-Akt-infected Int5 cells exhibited increased levels of peripherin when cultured in differentiation medium (Fig. 5D). Increased levels of peripherin are the result of Int5 cell differentiation but not of a direct effect of AAA-Akt, because peripherin levels of PC12 and Int5 cells in growth medium were not enhanced by Ad.AAA-Akt infection (Fig. 5D, growth). These results demonstrated that inhibition of Akt activity reversed the Wnt-dependent phenotype of Int5 cells, reinforcing the evidence that Akt has an important role in the Wnt signaling pathway. DISCUSSION Here we show that Wnt signaling stimulates Akt and that activated Akt, in association with Dvl, enhances the phosphorylation of GSK3␤ in the Axin complex. The concerted action of Dvl and Akt is a distinctive feature of Wnt signaling, which targets phosphorylation of the GSK3␤ pool that is directly responsible for phosphorylation of ␤-catenin and its subsequent rapid degradation. This finding explains why other factors, such as insulin and epidermal growth factor, that stimulate Akt through the PI3K pathway but do not activate Dvl fail to cause an increase in cytosolic ␤-catenin (30,31). We believe that Dvl interaction with Axin induces a conformational change, allowing Axin-bound GSK3␤ to be phosphorylated by Akt. This phosphorylated GSK3␤ cannot modify ␤-catenin efficiently, which consequently accumulates in the cytosol and translocates to the nucleus where it combines with LEF/TCF family members to upregulate Wnt-dependent gene expression.
Past reports have suggested that Akt is not involved in Wnt signaling (30,32). Similar to these accounts, we observed that Akt by itself was not sufficient to mimic a Wnt signal. Also consistent with this literature, constitutively active membranebound forms of Akt (14) did not enhance TOPFlash activity, even in the presence of Dvl overexpression (data not shown). Presumably, membrane-anchored Akt is unable to interact with the Axin complex in the cytosol and consequently cannot regulate GSK3␤ activity in this critical pool. In contrast to these constitutively active forms, wild type Akt is activated at the plasma membrane and then translocates to the cellular interior (15). Previous studies did not detect Akt activation and FIG. 5. Dominant-negative Akt restores neuronal differentiation of Int5 cells. A, Int5 cells were incubated in differentiation medium including vehicle (control) or sFRP-1 (15 g/ml) for 4 days. Cells were observed using phase contrast microscopy (original magnification, 200ϫ). B, Int5 cells were infected with control Ad.␤-gal.EGFP (left panels) or Ad.AAA-Akt (right panels) at MOI 50 for 2 , and then were maintained for 3 days in the differentiation medium. Cells were observed using phase contrast (upper panels) and fluorescence microscopy (lower panels) (original magnification, 400ϫ). Cells infected by the adenovirus were identified by GFP expression. C, The number of neurite-bearing cells compared with total cells is indicated as % differentiation. Data are presented as the mean Ϯ S.D. *, p Ͻ 0.001 versus Ad.␤-gal.EGFP. D, PC12 and Int5 cells were infected with control Ad.␤-gal.EGFP or Ad.AAA-Akt, and cells were maintained in growth medium or differentiation medium for 3 days. Western blots were probed with a peripherin antibody. To confirm equal protein loading, the same blot was also probed with an ␣-tubulin antibody.
resulting GSK3␤ phosphorylation at Ser 9 when cells were treated with Wnt/Wg-conditioned medium for short time periods, whereas free ␤-catenin was accumulated immediately by this treatment (8,31). In contrast, we documented Akt activation and involvement in Wnt signaling using models characterized by either prolonged or constitutive Wnt stimulation. Our data may suggest that the contribution of Akt is to sustain or enhance, but not initiate, the signaling process.
It is still unclear how Akt is activated by a prolonged Wnt stimulus. Recently, integrin-linked kinase was reported to interact with activated Dvl (33), to phosphorylate Akt (34), and to induce Wnt target genes (34,35). Future studies will determine whether Dvl activates Akt through the integrin-linked kinase pathway.