Endogenous Protein Kinase CK2 Participates in Wnt Signaling in Mammary Epithelial Cells*

Protein kinase CK2 (formerly casein kinase II) is a serine/threonine kinase overexpressed in many human tumors, transformed cell lines, and rapidly proliferating tissues. Recent data have shown that many cancers involve inappropriate reactivation of Wnt signaling through ectopic expression of Wnts themselves, as has been seen in a number of human breast cancers, or through mutation of intermediates in the Wnt pathway, such as adenomatous polyposis coli or β-catenin, as described in colon and other cancers. Wnts are secreted factors that are important in embryonic development, but overexpression of certain Wnts, such as Wnt-1, leads to proliferation and transformation of cells. We report that upon stable transfection ofWnt-1 into the mouse mammary epithelial cell line C57MG, morphological changes and increased proliferation are accompanied by increased levels of CK2, as well as of β-catenin. CK2 and β-catenin co-precipitate with the Dvl proteins, which are Wnt signaling intermediates. A major phosphoprotein of the size of β-catenin appears in in vitro kinase reactions performed on the Dvl immunoprecipitates. In vitro translated β-catenin, Dvl-2, and Dvl-3 are phosphorylated by CK2. The selective CK2 inhibitor apigenin blocks proliferation of Wnt-1-transfected cells, abrogates phosphorylation of β-catenin, and reduces β-catenin and Dvl protein levels. These results demonstrate that endogenous CK2 is a positive regulator of Wnt signaling and growth of mammary epithelial cells.

Wnt-1 was first identified as a proto-oncogene activated in mammary carcinomas caused by mouse mammary tumor virus and soon after was identified as the mammalian homolog of the Drosophila wingless gene (1,2). Transgenic mice overexpressing Wnt-1 develop mammary hyperplasia and carcinomas (3). Wnt-1 is one of the 16 known Wnt genes that encode secreted glycopeptide growth factors or morphogens (4). When expressed in C57MG, a normal mouse mammary epithelial cell line, many Wnts cause morphological transformation (5)(6)(7). Normal C57MG cells that reach confluence in culture appear as a monolayer of regular cuboidal cells; however, C57MG cells transfected with Wnt-1 have an elongated refractile morphology.
Protein kinase CK2 (formerly casein kinase II) has recently been added to the cascade of proteins in the Drosophila wingless pathway. In Schneider 2 cells, it was shown that CK2 associates with and phosphorylates Dsh (39). Transfection of mammalian Dvl-1 and Dvl-2 into clone-8 imaginal disc cell line results in hyperphosphorylation of Dvl proteins by CK2 (40). The Drosophila model suggests that CK2 could be an important regulator of Wnt signaling in other systems.
CK2 is a serine/threonine kinase that is ubiquitously expressed in both the cytoplasm and nucleus of eukaryotic cells. It exists as a constitutively active tetramer that contains two catalytic subunits, ␣ or ␣Ј (37-44 kDa), and two regulatory ␤ subunits (24 -28 kDa) (41,42). The two catalytic subunits are highly homologous, but the ␣Ј subunit has a unique required role in spermatogenesis (43). CK2 phosphorylates serines or threonines in acidic domains, with (S/T)XX(D/E) being the canonical motif (44 -47). CK2 was postulated to contribute to tumorigenesis because its activity is enhanced in many human solid tumors, transformed cell lines, and rapidly proliferating tissues (48 -51). Dysregulated expression of CK2 in cells can be oncogenic, as transgenic expression of CK2␣ can promote lymphoma (52)(53)(54) and breast cancer. 2 CK2 is characterized by the following biochemical properties: it is activated by polyamines (56 -58), inhibited by apigenin (chrysin) and 6-dichloro-1-␤-D-ribofuranosylbenzimidazole (59,60), and can utilize GTP as well as ATP as a phosphate donor (61). CK2 has been implicated in the regulation of many cellular processes, including DNA replication, basal and inducible transcription, and the regulation of cell growth and metabolism (62)(63)(64).
Here, we demonstrate that CK2 is up-regulated by the presence of a Wnt-1 signal in mouse mammary epithelial cells and that endogenous CK2 in mammalian cells associates with the Dvl proteins. Furthermore, ␤-catenin can also be found in this complex and is phosphorylated by CK2. Inhibition of CK2 activity by apigenin accelerates the degradation of ␤-catenin and Dvl proteins and causes cell cycle arrest. These results demonstrate that CK2 is a key regulator of the Wnt signaling pathway in mammalian cells.

MATERIALS AND METHODS
Cell Culture-pMV7-Wnt-1 (kindly provided by Dr. Anthony M. C. Brown) and control pMV7 empty vector were transfected into C57MG cells via electroporation. Stable clones were generated by selection in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2% L-glutamine, 1% penicillin/streptomycin (Cellgro, Mediatech Inc., Herndon, VA), 10 g/ml insulin (Sigma), and 400 g/ml G418 (Life Technologies, Inc.). Clones were screened for Wnt-1 expression, and results for several Wnt-1-positive clones were compared with those for neomycin-resistant clones transfected with the empty vector in each experiment. In addition to the stably transfected clones, we also analyzed pools of cells transiently infected with retrovirus expressing Wnt-1 to confirm our results. Retrovirally mediated gene transfer was performed essentially as described (65) with the following modifications: Phoenix cells (obtained from Dr. Gary Nolan) were plated at 1 ϫ 10 6 cells/60-mm dish and transfected overnight using the CLONTECH calcium phosphate transfection kit. The pBabe retroviral vector (66) and pBabe-Wnt-1 expression vector (a generous gift from Brian Elenbaas) were used at a concentration of 6 g/400 l of calcium phosphate solution. C57MG cells were plated at 5 ϫ 10 5 cells/60-mm dish 1 day prior to infection. 24 h after infection, C57MG cells were split and maintained in medium containing 2 g/ml puromycin.
Proliferation and Cell Cycle Analyses-To measure changes in mammary epithelial cell growth, Wnt-1-transfected clones were cultured with varying concentrations of apigenin (Sigma) for up to 2 days in microtiter wells. At 24 and 48 h, cells were treated with 50 l of freshly made 1 mg/ml 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-5-[(phenylamino)carbonyl]-2H-tetrazolium hydroxide (Polyscience Inc., Warrington, PA) and 25 M phenazine methosulfate (Aldrich) solution for 4 h at 37°C, at which time, the absorbance at 450 nm was determined with subtraction of background absorbance at 650 nm, providing quantitation of cell number (67). The absorbance at 450 -650 nm for medium alone was subtracted from each value. To determine the mechanism by which apigenin inhibits proliferation of the cells, the effects on cell cycle progression were determined. Cells were treated with apigenin overnight, harvested, resuspended in Nicoletti buffer (0.1% Triton X-100 and 0.1% sodium citrate) containing 0.5 mg/ml propidium iodide (Sigma), and analyzed on a flow cytometer (Becton Dickinson, Mountain View, CA) using the Cellquest program.
Northern Analysis-RNA was isolated using Ultraspec RNA (Biotecx Laboratories Inc. Houston, TX) and quantitated by spectrophotometer. 10 g of RNA was loaded onto a formaldehyde-based 1% agarose gel, electrophoresed, and blotted onto a nylon membrane (Gene screen Plus, NEN Life Science Products). The membrane was baked for 2 h at 80°C and prehybridized in Church buffer (7% SDS, 1% bovine serum albumin, 1 mM EDTA, 0.25M Na 2 HPO 4 , 0.17% H 3 PO 4 ). Full-length Wnt-1 and CK2 cDNAs were labeled with Klenow DNA polymerase (New England Biolabs Inc., Beverly, MA) and used to screen for Wnt-1 and CK2 expression.
Immunoprecipitation and Western Blotting-To extract cellular proteins, Wnt-1-expressing cells were lysed in buffer supplemented with protease and phosphatase inhibitors (50 mM Tris-HCl, pH 8.0, 1% Nonidet P-40, 125 mM NaCl, 1 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml pepstatin, 1 g/ml leupeptin, 1 mM Na 3 VO 4 , and 10 mM sodium pyrophosphate). The BCA protein assay (Pierce) was used to determine the protein concentrations of the whole cell lysates. Equal amounts of protein were loaded onto polyacrylamide gels, electrophoresed, and transferred onto nitrocellulose membranes (Schleicher & Schuell Inc., Keene, NH) using a semidry electroblotter (Owl Scientific, Woburn, MA). After transfer, the membranes were blocked overnight in 5% dry milk and probed for protein expressions using antibodies against CK2␣ (Upstate Biotechnology Inc., Lake Placid, NY), GSK3␤ (Transduction Laboratories, Lexington, KY), or ␤-catenin (Transduction Laboratories). Dvl proteins were detected with monoclonal antibodies: The monoclonal antibody to Dvl-1 has been described (68). Mouse monoclonals against Dvl-2 and 3 were generated against glutathione S-transferase fusion proteins containing either the C-terminal amino acids of Dvl-2 (2-10B5) or Dvl-3 (3-4D3) using the ClonaCell-HY hybridoma cloning kit (StemCell Technologies, Inc., Vancouver, British Columbia, Canada). The specificity of the three Dvl monoclonal antibodies was confirmed by Western blotting of extracts of COS cells transiently transfected with Dvl-1, Dvl-2, or Dvl-3 expression constructs, in which each antibody recognized its cognate Dvl only (data not shown). Immunoreactive bands using appropriate secondary antibodies coupled to horseradish peroxidase (Santa Cruz Biotechnology, Santa Cruz, CA) were visualized by chemiluminescence in signaling solution (Pierce). Ponceau S (Sigma) staining and/or a monoclonal ␤-actin antibody (Sigma) were used to confirm equal loading of all Western blot membranes.
For immunoprecipitation, 300 g of protein extracts was precleared with protein A-agarose beads (Sigma). 100 l of Dvl-1, Dvl-2, or Dvl-3 hybridoma supernatant or 1 l of 68 mg/ml ␤-catenin antibody (Sigma) was added to precleared lysates and incubated overnight at 4°C. 40 l of protein A-agarose beads was added to each of the incubations for 2 h at 4°C. Control immunoprecipitations with irrelevant antibody or beads alone were also performed. The immunoprecipitated complexes were washed five times in phosphate-buffered saline, and Western blotting was performed as described.
CK2 Kinase Assay-12 g of protein was incubated in buffer (100 mM Tris, pH 8.0, 20 mM MgCl 2 , 100 mM NaCl, 50 mM KCl, and 100 M Na 3 VO 4 ) with 5 Ci of [␥-32 P]GTP and 1 g/l CK2 substrate peptide (RRREEETEEE, Promega, Madison, WI) for 15 min at 30°C (69). Control kinase reactions without the peptide were also done for each of the samples. After 15 min of incubation, the kinase assays were stopped with 10 mM cold ATP and 0.4 N HCl. The samples were spotted onto a P81 Whatman filter circle and washed four times, 5 min each, with 150 mM H 3 PO 4 to elute unincorporated counts. Incorporated counts were quantified in an automatic scintillation counter. The samples were assayed in triplicate.
In Vitro Kinase Assay-Following immunoprecipitation of the whole cell lysates with Dvl-1, Dvl-2, and Dvl-3 antibodies as described above, the bead-bound proteins were incubated with 5 Ci of [␥-32 P]GTP at 37°C for 20 min alone or in the presence of apigenin (80 M) or wortmannin (25 nM). The kinase reactions were stopped by the addition of 2ϫ sample loading buffer. The samples were boiled, centrifuged, and subsequently loaded onto a denaturing SDS-polyacrylamide gel. The gel was transferred onto a nitrocellulose membrane and autoradiography of the membrane was performed to visualize [␥-32 P]GTP-labeled proteins. Western blotting was performed to identify the phosphorylated target using specific antibodies.
In Vitro Translation-pCI-␤-catenin (kindly provided by Dr. Bert Vogelstein), pSVK-Dvl-1, pBSK-Dvl-2, and pSVK-Dvl-3 plasmid DNA (1 g each) were in vitro transcribed and translated using the TNTcoupled reticulocyte lysate system (Promega). Plasmid DNAs were incubated for 75 min with appropriate polymerase in the presence of 20 Ci of [ 35 S]methionine. A negative control (no DNA) was prepared in the same manner. After the coupled transcription and translation reaction, 35 S-labeled protein products were calf intestinal phosphatasetreated as described (70) and immunoprecipitated with respective antibodies. These protein products were incubated with 10 units of recombinant CK2 enzyme (New England Biolabs Inc.) in the presence of 5 Ci of [␥-32 P]GTP at 37°C for 20 min. Products were loaded onto an acrylamide gel and transferred onto a nitrocellulose membrane. The phosphorylated proteins were visualized by autoradiography. The same membrane was cut, and Western blotting analysis was done to confirm the presence of the translated proteins.

Wnt-1 Induces
Up-regulation of CK2 and ␤-Catenin-Wnt-1expressing subclones of C57MG cells were generated to study the role of CK2 in Wnt signaling. C57MG cells were transfected with the pMV7 vector alone or pMV7-Wnt-1, and clones were selected in G418. Northern blotting (Fig. 1A) and reverse transcription-polymerase chain reaction were employed to identify Wnt-1 mRNA positive clones. Morphological differences between the Wnt-1-expressing clones and the control neomycintransfected clones were observed, as previously reported (5), and the proliferation rate of the Wnt-1 transfectants was uniformly greater than that of the controls. One high expressing clone was studied in detail, and the results were confirmed in other clones and in pools of cells infected with pBABE-Wnt-1 retrovirus. Northern analysis demonstrated that CK2 mRNA expression was elevated 2.5-fold in the Wnt-1-expressing clones (Fig. 1B); two different specific transcripts for CK2 are typically seen. CK2 activity was also up-regulated in the Wnt-1 transfectants, as determined using a synthetic CK2 peptide substrate and [␥-32 P]GTP as a phosphate donor. Wnt-1-expressing cells exhibited 2-fold higher CK2 activity than the vector controls (Fig. 1D). As expected in Wnt-1-overexpressing cells, ␤-catenin protein levels were up-regulated, as were CK2 protein levels. These results were seen in both the pMV7-Wnt-1-transfected and pBABE-Wnt-1-infected cells (Fig. 2).
To determine the functional importance of CK2 activity upon growth of the Wnt-1-transfected cells, we used the selective CK2 inhibitor apigenin. Apigenin rapidly inhibited CK2 activity in a dose-dependent manner (Fig. 3A) and dramatically blocked cell proliferation over the course of a 2-day treatment (Fig. 3B). [ 3 H]Thymidine incorporation was inhibited (not shown), and cells treated with apigenin accumulated in the G 2 /M phase of the cell cycle (Fig. 3C). Short term treatment with apigenin did not alter the morphology of the Wnt-1-expressing cells.
Endogenous CK2 and ␤-Catenin Associate with Dvl-To determine whether CK2 associates with Dvl in Wnt-1-expressing cells, immunoprecipitations with specific monoclonal antibodies raised against Dvl-1, Dvl-2, and Dvl-3 were performed, followed by Western blotting. Lysates prepared from whole cells were Western blotted with the different primary antibodies to ascertain the electrophoretic mobilities of the corresponding proteins (Fig. 4A). Each of the Dvl proteins was detected in Western blots of immunoprecipitates performed with the cognate antibody. In addition, we detected immunoreactive CK2 and ␤-catenin protein in the complexes precipitated by each of the Dvl antibodies (Fig. 4B). GSK3␤ was not detected in the Dvl immunoprecipitates. Control immunoprecipitates using nonspecific antibodies did not precipitate any of the proteins. We attempted to perform the reciprocal immunoprecipitates; however, the CK2 antibody failed to work for immunoprecipitation. The ␤-catenin antibody was capable of inefficiently immunoprecipitating a minor fraction of the ␤-catenin in the cell lysate, but neither CK2 nor Dvl proteins could be detected in this material. This may be a quantitative problem, because most ␤-catenin is involved in the formation of adhesion complexes with E-cadherin; alternatively, the polyclonal antibody may preferentially precipitate monomeric ␤-catenin, or in fact it may compete with CK2 and Dvl for binding to the ␤-catenin.
␤-Catenin Is Phosphorylated by CK2 in Vitro-To determine whether the CK2 in the Dvl immunoprecipitates was capable of phosphorylating its partners, an in vitro kinase reaction was performed on the immunoprecipitates. We found that a single major phosphoprotein of approximately 97 kDa appeared in all three Dvl immunoprecipitates (Fig. 5A). This protein migrated in polyacrylamide gels slightly more slowly than immunoreactive ␤-catenin detected on Western blotting of the same gel, consistent with the phosphorylated form of ␤-catenin (71). Dvl-1 migrates slightly more slowly than the 97-kDa phosphoprotein, whereas Dvl-2 and Dvl-3 run faster. These data suggested that the phosphoprotein might be ␤-catenin; the fact that the 97-kDa phosphoprotein appeared in reactions employing [␥-32 P]GTP as a phosphate donor supports the hypothesis that CK2 is phosphorylating it directly, as CK2 is one of the few cellular kinases that can employ GTP as a substrate (61). Moreover, the CK2 inhibitor apigenin completely abolished phosphorylation of the 97-kDa band in the in vitro kinase reaction (Fig. 5A). Apigenin is a plant flavonoid that is selective for CK2 (60). One report suggests that it may also inhibit the lipid kinase PI3K in vitro (72); however, wortmannin, an authentic PI3K inhibitor, did not affect phosphorylation of the 97-kDa phosphoprotein band (Fig. 5B), nor is there any evidence that PI3K can utilize GTP as a phosphate donor. Immunodepletion of ␤-catenin using four rounds of immunoprecipitation with the ␤-catenin antibody completely abrogated this phosphorylated band, providing further evidence that the substrate is ␤-catenin (Fig. 5C). As previously stated, immunoprecipitation of ␤-catenin by polyclonal ␤-catenin antibody did not bring down the CK2 or Dvl proteins. Thus, in vitro kinase reaction of ␤-catenin immunoprecipitate alone could not generate a phosphoprotein as expected. Although the ␤-catenin immunoprecipitates do not contain CK2 or Dvl, treatment of the immunoprecipitates with exogenous recombinant CK2 results in the generation of a phosphoprotein of identical mobility (Fig. 5D). Finally, in vitro transcribed and translated ␤-catenin can also be phosphorylated by CK2, as can Dvl-2 and Dvl-3 (Fig. 5E). Dvl-1 was not efficiently transcribed and translated, so we could not definitively determine whether or not it can be phosphorylated. ␤-Catenin has multiple CK2 consensus phosphorylation sites, including one at serine 29, in close proximity to the critical GSK3 phosphorylation site at serine 37 that signals for ubiquitination and degradation (19).
Apigenin Causes Destabilization of ␤-Catenin and Dvl Proteins-Having shown that apigenin treatment blocks phosphoexpressing cells were treated with 40 M apigenin overnight. The cells were harvested and stained with propidium iodide, and relative DNA content was analyzed by fluorescence-activated cell sorting.

FIG. 3. Apigenin inhibits CK2 activity in a dose-dependent manner, inhibits cell proliferation, and causes a G 2 /M arrest.
A, Wnt-1-expressing cells were untreated (Un Rx) or treated with increasing concentrations of apigenin (Api) for 2 h, and lysates were prepared and assayed for CK2 activity as described in Fig. 1D. B, Wnt-1-expressing cells in microtiter wells were treated with apigenin. At 24 and 48 h, an assay with 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-5-[(phenylamino)carbonyl]-2H-tetrazolium hydroxide was performed to measure cell number. Absorbance at 450 -650 nm is plotted for end point readings at 4 h, with subtraction of values for medium alone. Samples were assayed in quadruplicate, and results are expressed as mean Ϯ S.D. C, Wnt-1-

FIG. 4. CK2 and ␤-catenin associate with Dvl-1, Dvl-2, and
Dvl-3. A, 100 g of protein extracted from Wnt-1-expressing cells was run on a denaturing polyacrylamide gel and Western blotted with antibodies to ␤-catenin and Dvl-1, Dvl-2, and Dvl-3 to ascertain the electrophoretic mobilities of the corresponding proteins. B, 300 g of precleared protein lysates from Wnt-1-transfected cells was incubated overnight with the specific Dvl monoclonal antibodies followed by a 2-h incubation with protein A-agarose beads, which were washed thoroughly, boiled in 2ϫ sample buffer, and applied to a polyacrylamide gel. A control (C) immunoprecipitation (IP) was performed with an unrelated mouse monoclonal antibody. By Western blotting, each Dvl antibody immunoprecipitated its cognate Dvl. ␤-Catenin and CK2 were found to co-immunoprecipitate with the Dvl proteins. GSK3␤ was not detected in the Dvl immunocomplexes; the band migrating slightly more slowly than the GSK3␤ protein detected in the whole cell lysate is the immunoglobulin heavy chain (IgH chain). 5. CK2 phosphorylates ␤-catenin. A, an in vitro kinase reaction was performed on thoroughly washed Dvl-1, Dvl-2, or Dvl-3 immunoprecipitates using [␥-32 P]GTP as a phosphate donor. The reaction was stopped by the addition of 2ϫ sample buffer, electrophoresed, and transferred onto a nitrocellulose membrane. The [␥-32 P]GTP labeled proteins were identified by autoradiography (left panel). Western blotting (WB) of the same membrane was subsequently performed with the ␤-catenin antibody (right panel). A single major phosphoprotein (upper arrow) migrating slightly more slowly than ␤-catenin (lower arrow) was seen in the in vitro kinase (IVK) reaction. Phosphorylation of this protein was completely abolished in the presence of 80 M apigenin (Dvl-1ϩApi). Control immunoprecipitates (Cont) lacked both the phosphoprotein and ␤-catenin. B, because CK2 is one of the few kinases capable of utilizing GTP as a phosphate donor, the apigenin-inhibitable enzyme active in the in vitro kinase reaction was likely to be CK2. Because apigenin can also inhibit PI3-kinase in vitro, we used an authentic PI3K inhibitor, wortmannin, and found no decrease in the phosphoprotein band. Western blot with ␤-catenin antibody (bottom panel) documents the presence of ␤-catenin protein in each immunoprecipitate. C, to further confirm the identity of the phosphoprotein as ␤-catenin, lysates were subjected to immunodepletion with the ␤-catenin antibody prior to Dvl-1 immunoprecipitation and the in vitro kinase reaction. No residual phosphoprotein band was seen. D, when recombinant CK2 was added to the ␤-catenin immunoprecipitate, a phosphoprotein having same mobility as endogenously phosphorylated phosphoprotein seen in the Dvl immunoprecipitate was seen. E, in vitro translated ␤-catenin, Dvl-2, and Dvl-3 are phosphorylated by recombinant CK2. To further confirm that ␤-catenin is a substrate of CK2, ␤-catenin, Dvl-2, and Dvl-3 plasmids were subjected to in vitro transcription and translation. The protein products were treated with calf intestinal phosphatase, immunoprecipitated with their respective antibodies, and subjected to an in vitro kinase reaction with [␥-32 P]GTP in the presence or absence of recombinant CK2. Translated ␤-catenin, Dvl-2, and Dvl-3 were phosphorylated by CK2 (upper panel). Western blotting of the same membranes confirmed that the proteins were translated successfully (lower panels). Arrows identify the bands of identical mobility in the two autoradiographs as ␤-catenin (regular arrow), Dvl-2 (round-headed arrow), and Dvl-3 (triangular-headed arrow). rylation of the 97-kDa phosphoprotein in an in vitro kinase reaction, we treated cells with apigenin to determine whether this would affect the stability of ␤-catenin and Dvl proteins in vivo. In a dose-dependent manner, we found a dramatic decrease in the levels of ␤-catenin and Dvl-1, Dvl-2, and Dvl-3 (Fig. 6). Apigenin has also been described as inhibiting MAP kinase (73), although this was in whole cells only; in vitro experiments show no inhibition of MAP kinase. 3 Accordingly, neither the PI3K inhibitor wortmannin nor the MAP kinase inhibitor PD98059 altered the levels of ␤-catenin or Dvl-1, Dvl-2, or Dvl-3 proteins. To determine whether the reduction in ␤-catenin occurred through a decreased rate of synthesis or increased rate of degradation, we measured the half-life of the protein in the presence of cycloheximide that blocked new protein synthesis. We found that ␤-catenin is quite stable in Wnt-1-expressing cells, with a half-life of more than 5 h (Fig. 7), similar to what has been described in COS cells (35) and in C57MG cells (75,76). The Dvl proteins appear to be equally stable. However, in the presence of apigenin, protein levels rapidly decline. Immunoreactive Dvl disappears in less than 30 min, whereas ␤-catenin levels decline to less than 50% of baseline and then plateau, suggesting that not all pools of ␤-catenin are regulated via apigenin-sensitive phosphorylation.

DISCUSSION
In Wnt-1-transfected or -infected C57MG mouse mammary epithelial cells, CK2 is up-regulated at the level of mRNA, protein, and activity. As expected, Wnt-1-transfected cells also up-regulate ␤-catenin. To determine whether endogenous CK2 associates with Dvl in Wnt-1-expressing mammalian cells, as occurs in insect cells with overexpression of Dvl or Drosophila frizzled 2, immunoprecipitation experiments were performed using specific monoclonal antibodies raised against the Dvl proteins. For each antibody, the cognate Dvl and also CK2 were detected by Western blot. GSK3␤ was not detected in the complex, but unexpectedly, ␤-catenin protein was consistently found in these immunoprecipitates. When [␥-32 P]GTP was provided in an in vitro kinase reaction, a single major 97-kDa phosphoprotein migrating slightly above the ␤-catenin protein was seen in common in all the Dvl immunoprecipitates (but not in the controls). In addition to the fact that CK2 is one of the few kinases capable of utilizing GTP as a phosphate donor, further evidence that CK2 was responsible for generation of this phosphorylated band was the fact that phosphorylation was inhibited by apigenin, a selective CK2 inhibitor. The same band was also seen in CK2-phosphorylated ␤-catenin immunoprecipitates and in in vitro translated ␤-catenin treated with recombinant CK2; this band disappeared with immunodepletion of ␤-catenin. We also find that in vitro translated Dvl-2 and Dvl-3 are substrates for CK2, consistent with recent reports about Drosophila dsh and mammalian Dvl-1 and Dvl-2 (40). These experiments indicate that ␤-catenin not only associates with Dvl proteins and with CK2 but is a target of CK2 phosphorylation.
This association appears to be quite important functionally, as CK2 activity is essential for maintenance of ␤-catenin and Dvl protein levels in the mammary epithelial cells, and for their proliferation. Treatment of cells with apigenin reduces CK2 activity as expected, dramatically accelerates the degradation of ␤-catenin and Dvl, and leads to a G 2 /M cell cycle arrest. Although apigenin has been reputed to also inhibit PI3K and MAP kinase (although these observations were carried out after treatment of live cells, rather than isolated enzymes), the authentic PI3K and MAP kinase inhibitors wortmannin and PD98059 do not have the same effects. The Dvl proteins have multiple CK2 consensus phosphorylation sites, as does ␤-catenin. In ␤-catenin, strong consensus sites at serine 29 and threonine 59/serine 60 flank the serine 37 that is the GSK3␤ consensus site critical for initiating ubiquitination and degradation. CK2 phosphorylation of ␤-catenin may render it resistant to degradation, or the association of CK2 with ␤-catenin may block access by GSK3␤ and the ubiquitination machinery.
Recently, casein kinase I CK1 was identified in an expression screen for mRNAs capable of causing axis duplication in Xenopus oocytes (77). Xenopus CK1 bound Xenopus dishevelled (Xdsh) in a yeast two-hybrid assay, and injection of CK1 mRNA in oocytes augmented Xdsh phosphorylation but did not result in phosphorylation of ␤-catenin. Thus, CK1 also could potentially affect Wnt signaling in mammalian cells, although this has not yet been demonstrated. This would be an unexpected finding, because CK1 and CK2 are not closely related enzymes and generally have no overlapping functions. Although both enzymes are serine-threonine kinases that can phosphorylate casein, neither is a true casein kinase in vivo, and their consensus phosphorylation sites are distinct. CK2 is a biochemically very different monomeric enzyme that is incapable of 3 T. Means, and M. Fenton, unpublished data. utilizing GTP as a phosphate donor and is not inhibited by apigenin.
Our data are consistent with the model in which ␤-catenin participates in several different pools in cells, as indicated by biochemical studies in which complexes of different sizes can be separated by column chromatography (76,78,79). One major pool is the transmembrane calcium-dependent homotypic adhesion complex in which ␤-catenin complexes with E-cadherin and other proteins (Fig. 8). When ␤-catenin participates in Wnt-1 signaling, genetic and biochemical data have identified a complex in the cytoplasm consisting of ␤-catenin, APC, axin or axil, Dvl, Frat1 (80), and GSK3␤ that regulates the rate of ubiquitination and proteosomal degradation of ␤-catenin. We have termed this the "negative regulatory complex," as GSK3␤ is active in the absence of a Wnt signal, and mutation of APC or ␤-catenin leads to the dissociation of the complex and stabilization of ␤-catenin. Our data are consistent with the existence of another cytoplasmic complex consisting of ␤-catenin, Dvl, and CK2 that has a "positive regulatory" role. It is enhanced by Wnt-1 signaling, and inhibition of CK2 markedly destabilizes ␤-catenin and Dvl and blocks cell proliferation. We did not detect GSK3␤ in the Dvl/CK2 complex, so our evidence suggests these two cytoplasmic complexes are distinct, but we cannot definitively rule out the possibility that some complexes involve Dvl, CK2, ␤-catenin, GSK3␤, APC, and axin all together. Recently, two phosphatases, PP2A and PP2C, have also been implicated in the Wnt signaling cascade. The catalytic subunit of PP2A dephosphorylates axin and stabilizes ␤-catenin, whereas overexpression of the B56 regulatory subunit binds to APC and down-regulates ␤-catenin (74,81). PP2C binds to Dvl, ␤-catenin, and axin and up-regulates LEF-1-dependent transcription through down-regulation of axin (55). Ultimately, through a balance of complex positive and negative regulatory influences upon ␤-catenin levels in the cytoplasm, ␤-catenin becomes free to translocate into the nucleus, where it associates with TCF/LEF to activate transcription of target genes, such as myc and cyclin D1.
This intricacy of signaling complexes involving ␤-catenin is an almost unprecedented mode of regulation, but it may reflect the multifunctional nature of ␤-catenin and its critical role in both adhesive and signaling processes. Given the elaborateness of ␤-catenin regulation, it is no surprise that stabilization of ␤-catenin levels in cells through Wnt overexpression or by mutation of ␤-catenin itself or of APC can cause tumorigenesis in humans or in animal models. One might also hypothesize that some tumors might occur through mutation of axin, through up-regulation of CK2 activity, or through down-regulation or loss of GSK3 (i.e. GSK3 might be an "anti-oncogene"). We and others have in fact shown that CK2 is up-regulated in many human tumors, and we have demonstrated that transgenic overexpression of CK2 promotes tumorigenesis in mice. Preliminary experiments suggest that in CK2␣ transgenic tumors ␤-catenin is up-regulated, although CK2 has many other targets in cells, and this may not be its only mechanism of CK2-dependent transformation.
Acknowledgments-We thank Dr. Anthony M. C. Brown for kindly providing the pMV7-Wnt-1 construct and Dr. Bert Vogelstein for pro-

FIG. 8. ␤-Catenin participates in multiple complexes in cells.
Our data and data in the literature are consistent with a model in which ␤-catenin (␤cat) in cells partitions into several distinct pools. ␤-Catenin is a co-factor in calcium-dependent homotypic adhesion complexes with E-cadherin and other proteins, identified here as the membrane adhesion complex. Genetic and biochemical studies of Wnt and wg signaling have identified a negative regulatory complex, which contains APC, GSK3␤, Dvl, Frat1, and a bridging molecule, axin or axil. This complex controls ␤-catenin degradation via the proteosome pathway. Recent data have shown that phosphatases PP2A and PP2C dephosphorylate axin, thereby inhibiting the negative regulatory complex. Our data suggest that ␤-catenin can also participate in a positive regulatory complex with Dvl proteins and CK2, and CK2 activity promotes Wnt signaling. An equilibrium between the negative and positive regulatory complexes appears to determine the amount of ␤-catenin that is available for translocation to the nucleus to serve as a cofactor for high mobility group box transcription factors of the TCF/LEF family.