Protein kinase CKII regulates the interaction of β-catenin withα-catenin and its protein stability

β-Catenin was originally identified in association with the cell adhesion molecule E-cadherin (Vestweber and Kemler, 1984; Ozawa et al., 1989; Nagafuchi and Takeichi, 1989), but has also recently won increasing interest as an important mediator in the Wnt signaling pathway (for a review, seePolakis, 2000). In the cadherin cell-cell adhesion complex β-catenin plays a central role since it binds to both the cytoplasmic domain of E-cadherin and to α-catenin,which in turn binds directly to F-actin and other actin-associated proteins(Ozawa et al., 1990;Stappert and Kemler, 1994;Huber and Weis, 2001). This association with the actin-based cytoskeleton is important for the adhesion function of all classical cadherins, as best demonstrated for the E-cadherin-catenin complex in adherens junctions of epithelial cells (for a review, see Aberle et al.,1996a). Although the respective binding sites between E-cadherin and β-catenin and between α- and β-catenin have been characterized in detail (Stappert and Kemler, 1994; Aberle et al.,1996b; Huber et al.,1997; Pokutta and Weis,2000; Yang et al.,2001), recent evidence underlines the importance of post-translational modifications in modulating the function of this adhesion complex. Phosphorylation of E-cadherin by casein kinase II (CKII) and glycogen synthese kinase-3β (GSK-3β) increases the affinity of E-cadherin forβ-catenin and thus enhances E-cadherin-mediated cell adhesiveness(Huber and Weis, 2001;Lickert et al., 2000). In contrast, reduced cell adhesion has been reported upon tyrosine phosphorylation of β-catenin (Ozawa and Kemler, 1998; Hazan and Norton, 1998; Gaudry et al.,2001) and an association of β-catenin with receptor-tyrosine kinases, EGF-R, IGF-R and c-Met(Hoschützky et al., 1994;Hiscox and Jiang, 1999;Playford et al., 2000) as well as with c-src (Kinch et al.,1995), has been shown. In comparison, little is known about the role of serine/threonine (Ser/Thr) phosphorylation of β-catenin in modulating the function of the E-cadherin adhesion complex. It is known that phosphorylation of β-catenin by GSK-3β determines its availability as a transducer of Wnt signals. In the absence of Wnt, phosphorylation by GSK-3β targets β-catenin for degradation(Yost et al., 1996;Aberle et al., 1997;Orford et al., 1997). Phosphorylation of β-catenin by GSK-3β occurs in a multiprotein complex that includes the tumor suppressor Adenomatous Polyposis Coli (APC)(Munemitsu et al., 1995), Axin(Zeng et al., 1997), or Conductin (Behrens et al.,1998). The composition of this multiprotein complex and the integration of β-catenin depend on the phosphorylation of each component by GSK-3β (reviewed by Seidensticker and Behrens, 2000; Dominguez and Green, 2001). Wnt inhibits GSK-3β-mediated phosphorylation of β-catenin and under these conditions β-catenin enters the nucleus and forms complexes with transcription factors of the LEF/Tcf family, resulting in specific transcriptional activation (reviewed byRoose and Clevers, 1999;Hecht and Kemler, 2000).

β-catenin is implicated in quite different cellular processes, which requires a fine-tuned regulation of its function, so it is very likely thatβ-catenin is a substrate for other yet to be identified protein kinases. Indeed, it was recently reported that CKII phosphorylates β-catenin(Song et al., 2000), and that CKII and β-catenin co-immunoprecipitate with Dvl proteins, the mammalian homologues of Drosophila Dishevelled (Dsh). From these results it was concluded that CKII participates in Wnt signaling and may act as a positive regulator in this pathway although the underlying molecular mechanisms are at present poorly understood.

CKII exists as a constitutively active tetramer that contains two catalytic(α or α′) and two regulatory (β) subunits(Pinna and Meggio, 1997;Allende and Allende, 1995). Although more than 160 substrates have been identified to date, the regulation of this ubiquitously expressed pleiotrophopic kinase remains unclear. A nuclear shift of CKII-α′ during G₁-phase and in proliferating cells (Seldin and Leder,1995; Keliher et al.,1996; Landesman-Bollag et al.,1998; McKendrick et al.,1999; Ahmed, 1994)points towards a role of CKII in mitotic control and proliferation. However,due to the broad subcellular distribution, it is generally assumed, that CKII is controlled by different interaction partners and in different subcellular compartiments (for a review, see Faust and Montenarh, 2000).

In a search for protein kinases that use β-catenin as substrate we confirmed that CKII also phosphorylates β-catenin. We have now identified amino acid (aa) residues in β-catenin phosphorylated by CKII and performed a mutational analysis to obtain first insights into the biological function of this post-translational modification. By comparing wild-type (wt)and Ser/Thr-mutated (Ser/Thr-mutant) β-catenin in kinase assays in vitro and in vivo we provide evidence here that CKII regulates the best studied functions of β-catenin (i.e. its central role in the E-cadherin adhesion complex and its tight control of cytoplasmic stability), which is a prerequisite for the canonical Wnt signaling pathway.

Mouse epithelial carcinoma CMT cells (ATTC, CCL-223) and human kidney epithelial 293 cells (ATTC, CRL-1573) were grown at 37°C/10%CO₂ in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal calf serum (FCS) and 2 mM L-glutamine. 293 cells stably expressing either hemagglutinin (HA)-tagged wt or Ser/Thr-mutantβ-catenin were selected in the presence of 100 μg/ml Zeocin(Invitrogen).

Mouse monoclonal antibodies against α-catenin, β-catenin,Protein kinase A (PkA), and GSK-3β were obtained from Transduction Laboratories (Lexington, KY), rabbit polyclonal antibodies against CKII were from Santa Cruz Biotechnologies, and rat monoclonal antibodies against hemagglutinin (HA) (clone 3F10) were from Roche. Anti-GST antibodies were from Sigma, and rabbit polyclonal anti-Ki67 was from Novocastra Laboratories(Newcastle). Anti-myc antibodies (Evan et al., 1985) were purified from supernatants of hybridoma clone 9E10 grown in DMEM containing 3% FCS. Fluorescein-conjugated secondary antibodies were purchased from Dianova. 3000 mCi/ml [γ-³²P]ATP was from Amersham Pharmacia Biotech.

293 cells grown on collagen- or poly-L-lysine-coated coverslips were washed in PBS, pH 7.4, and fixed in 3% paraformaldehyde/PBS, pH 7.4 at room temperature (RT) for 20 minutes. Free aldehyde groups were blocked with 1 M glycine/PBS, pH 8.5, for 5 minutes. Cells were permeabilized with 0.5% Triton X-100 for 5 minutes. Incubation was performed with primary antibodies at 2μg/ml for 1 hour at 37°C and with fluorescein-conjugated secondary antibodies (Dianova) for 1 hour at 37°C in the dark. Cells were kept in a mounting solution (50% glycerol; 50% PBS; 100 μg/ml 1,4-diazabicyclo-[2,2,2]octone) and digital images were taken with a computer-controlled digital C4880 camera (Hamamatsu, Japan) on an Axioskop microscope (Zeiss, Jena). Camera and microscope were controlled by the computer program Openlab (Improvision, Coventry, UK).

Cells grown to 80% confluency were washed with PBS and lysed in 500 μl CSK buffer [150 mM NaCl, 10 mM PIPES (pH 6.8), 3 mM MgCl₂, 0.5%Triton X-100, 10 mM NaF, 1 mM sodium vanadate, 10 μg/ml leupeptin, 10μg/ml phenylmethylsulfonyl fluoride (PMSF)] on ice for 20 minutes. Cell debris was removed by centrifugation at 16,000 g for 10 minutes. The amount of total protein was measured using the BCA-Kit (Pierce)and equal amounts of total protein were used for each analysis. Immunoblot analysis was performed as described (Aberle et al., 1996b), but using PVDF membranes (Millipore). Immunoprecipitation experiments were done as described(Hoschützky et al., 1994)with the following modifications: lysis and precipitations were performed in CSK buffer and the immunocomplexes were incubated overnight at 4°C.

Sequential detergent extraction was performed as described(Ramsby and Makowski, 1999). Briefly, after washing the cells three times in PBS, cytosolic and soluble cytoskeletal proteins were released with 0.015% digitonin, prior to the extraction of membrane and organelle-components with 0.5% Triton X-100. Nuclear and cytoskeletal proteins were subsequently solubilized with 1% SDS,prior to 10 fold dilution with normal CSK-buffer. All other steps were carried out on ice.

For transient transfection experiments 1×10⁶ 293 cells cultured in 9-cm culture dishes for 5-7 hours were incubated overnight with a calcium-phosphate coprecipitate containing 5 μg wt or Ser/Thr-mutantβ-catenin DNA; 4 μg DNA for mouse Axin-1 fused to a Myc tag; 2 μg Myc-GSK-3β DNA; 2 μg Myc-tagged CKII DNA, or 2 μg Myc-tagged ERK2. Cells were washed three times with PBS and cultured for 36 hours before analysis.

For pulse-chase experiments 2.5×10⁵ 293 cells stably expressing either wt or Ser/Thr-mutant β-catenin were cultured in 3.5-cm culture dishes for 6 hours. Cells were then washed twice with PBS, starved for 1 hour in cysteine- and methionine-free medium, incubated with[³⁵S]cysteine/[³⁵S]methionine (150 Ci/ml) for 1 hour,washed and chased for 5 hours. Cells were lysed in 100 μl CSK buffer containing 0.5% SDS and boiled for 10 minutes. The lysates where then diluted 10-fold and equal amounts of incorporated radioactivity were subjected to immunoprecipitation experiments. Precipitates were resolved by SDS-PAGE and the intensities of the products were quantified using the phospho-imager.

CMT cells grown to 80% confluency in 9-cm dishes were lysed in 500 μl CSK buffer and immunoprecipitations were usually done overnight at 4°C with anti-β-catenin antibodies coupled to protein A-Sepharose(Amersham-Pharmacia Biotech). Precipitates were washed three times in ice-cold PBS containing 0.5% Triton X-100, 150 mM NaCl, 10 mM NaF, 2 mM sodium molybdate, 1 mM sodium vanadate. Kinase activities present in the immunocomplexes were eluted by applying ascending salt concentrations(200-1000 mM NaCl) in distilled water containing phosphatase inhibitors (10 mM NaF, 10 mM sodium molybdate, 1 mM sodium vanadate). After brief centrifugation the supernatants were diluted to physiological salt concentrations (100 mM NaCl) and the volumes were adjusted to 400 μl. In control assays cell lysates were incubated with 1 μg mouse IgG of unrelated specificity.

For in vitro kinase assays with eluted kinase activity, the indicated GST fusion proteins immobilized on glutathione-Sepharose 4B (Amersham Pharmacia)were incubated with 100 μl of the eluate at RT for 30 minutes. After three washes with PBS containing 0.5% Triton X-100, 150 mM NaCl, 10 mM NaF, 2 mM sodium molybdate and 1 mM sodium vanadate, kinase assays were performed in 50μl kinase buffer [20 mM Tris-HCl (pH 7.4), 10 mM MgCl₂, 1 mM CaCl₂, 1 μM ATP] including phosphatase inhibitors and 7 μCi[γ-³²P]ATP at RT for 20 minutes. Reactions were stopped by adding 20 μl of stop buffer [10 mM EDTA (pH 8.0), 10 mM sodium phosphate(pH 8.0), 10 mM Na₄P₂O₇] and washed three times with cold PBS containing 0.5% Triton X-100, 1000 mM NaCl, 10 mM NaF, 2 mM sodium molybdate, 1 mM sodium vanadate. Radioactive gels were Coomassie Blue-stained and radioactivity was quantified with a BAS 1000 Bioimaging Analyzer (Fuji), while Coomassie Blue staining was quantified using the NIH Imager program version 1.59. CKII was specifically inhibited with the following peptides: (Arg)₃ Ala Asp Asp Ser (Asp)₅(competitor); and (Arg)₃ Ala Asp Asp Ala (Asp)₅(control).

To pre-phosphorylate recombinant β-catenin with CKII, 2 μg of GSTβ-catenin linked to GSH beads was incubated in CKII kinase buffer (20 mM Tris-HCl, 50 mM KCl, 10 mM MgCl₂, pH 7.5) with 20 mM non-radioactive ATP and 1 U of recombinant CKII (New England Biolabs) at 25°C for 20 minutes. Reactions were stopped by adding 50 μl stop-buffer(see above). To remove all residual kinase-activity, the reactions were extensively washed 5 times with 1 ml ice-cold PBS containing 1000 mM NaCl,0.5% Triton X-100 and phosphatase-inhibitors. The NaCl concentration was reduced by 2 additional washing steps with PBS containing 100 mM NaCl. Pre-phosphorylation with GSK-3β was done in the same way, but including 5 mM DTT in the assay-buffer, using 0.1 U of recombinant GSK-3β (New England Biolabs), and performing the reaction at 30°C.

All GST fusion proteins were expressed in E. coli BL21 (DE3). Affinity purification on GSH-Sepharose beads (Amersham Pharmacia) was carried out as described (Aberle et al.,1996b). Proteins were dialyzed against 20 mM HEPES (pH 8.0). Forβ-catenin fragments including only aa residues 302-535 or 535-683,concentration with Centricon ultrafiltration cartridges was required.

After purification from E. coli BL21 (Amersham Pharmacia Biotech)GST-Axin was loaded on a PD10 column to get rid of some major degradation products. The eluted and pooled fractions were concentrated as described above. β-catenin-His₆ was expressed in E. coli BL21(DE3) and affinity purified on Talon-Beads (Clontech) as described(Aberle et al., 1996b). All recombinant proteins were aliquoted and stored at -20°C.

If not indicated otherwise, 2 μg of GST-protein linked to GSH-beads was incubated with recombinant His₆-tagged protein pre-absorbed with GSH-beads. Reactions were carried out in 500 μl association-buffer (20 mM Tris-HCl, 10 mM MgCl₂, 0.1% Triton X-100, 100 mM KCl) at RT for 30 minutes. Protein complexes were collected by brief centrifugation (1 minute,800 g, 4°C), and washed three times with cold PBS containing 0.5% Triton X-100, 160 mM NaCl, 10 mM NaF, 2 mM sodium molybdate, 1 mM sodium vanadate.

GST expression constructs pGEX4T1MMBC, pGEX4T1BNTERM (coding forβ-catenin in residues 1-119) and pGEX4T1BCTERM (aa 683-781) were previously described (Aberle et al.,1996b). To obtain pGEX4T1B1-302, pSKβtot was digested with BamHI/BglII and the resulting 906 bp-fragment coding for the first 302 aa residues of β-catenin was cloned into BamHI-cleaved pGEX4T1. PGEX4T1B302-535 was obtained by cloning the 1176-bp BglII-fragment into BamHI-cleaved pGEX4T1. The resulting plasmid coding for residues 302-781 was digested with SpeI/XhoI and blunt-end ligated to obtain a construct coding for β-catenin missing aa residues 535-781.

For construction of pGEX4T1B535-683 a 741 bp-fragment coding for residues 535-781 was amplified from pGEX4T1MMBC by PCR, using the following primers:β535ff (5′-GCGGGATCCCGAC-TAGTTCAGCTG-3′) and βctermrev(5′-TTACAGGTCAGTAT-CAAACCAGGC-3′). The fragment was cloned into BamHI/EcoRV digested pGEX4T1BΔCTERM(Aberle et al., 1996b).

The plasmids coding for Myc-tagged mouse Axin 1 and Myc-tagged ERK2 where kindly provided by L. Zeng (Columbia University, New York, NY) and P. E. Shaw(The Beatson Institute for Cancer Research, Glasgow, UK).

For the generation of C-terminal HA-tagged β-catenin, oligonucleotides encoding the influenza virus HA-tag were inserted into pCS²⁺ after EcoRI restriction of the vector. cDNA coding for β-catenin aa residues 1-781 was then cloned in-frame in front of the HA-tag region.

Site-directed mutagenesis was performed as described(Higuchi, 1990). For all three mutants the same wt primers were used:β1ff:5′-ATGGCTACTCAAGCTGACC-3′, and β1875rev:5′-GGCCTCT-GCAGCCTCCTTGTCC-3′.

For generation of the single mutants the following mutagenic primer pairs were used: M1ff: 5′-CTGGCAGCAGCAGGCT-TACTTGGATTCTGG-3′ and M1rev:5′-CCAGAATCCAAGT-CAGCCTGCTGCTGCCAG-3′. M2: M2ff:5′-ATGTTCCCTGAG-GCTCTAGATGAGGGC-3′, and M2rev:5′-GCCCTCATCTAGAGC-CTCAGGGAACAT-3′. M3: M3ff:5′-GGCATGCAGATCCCATC-CCAGCAGTTTGACGCT-3′, and M3rev:5′-AGCGTCAAACTG-GTCGGATGGGATCTGCATGCC-3′.

After NcoI digestion, the PCR products were cloned into pCS²+β-catenin-HA. To generate the Ser/Thr-mutant,HA-β-catenin-M3 was first subcloned after into HA-β-catenin-M2 after SphI/SpeI digestion of the vector. A HincII/SpeI-fragment containing both mutations, M2 and M3,was then cloned into HA-β-catenin-M1. HA-tagged wt β-catenin and Ser/Thr-mutant constructs were cloned into the pBUD CE4 vector (Invitrogen)under control of the EF-1α promoter. For stable transfection of 293 cells the calcium-phosphate method was used.

It has been reported that GSK-3β(Yost et al., 1996;Aberle et al., 1997;Orford et al., 1997), and the protein kinase casein kinase II (CKII)(Song et al., 2000)phosphorylate β-catenin, although for the latter the exact Ser/Thr residues phosphorylated are not known. In a screen for Ser/Thr kinases that phosphorylate β-catenin, we also observed phosphorylation ofβ-catenin by CKII. In this approach, kinase activities associated withβ-catenin in vivo were coprecipitated with anti-β-catenin antibodies from cell lysates and were eluted with an increasing salt gradient. Each fraction was tested for phosphorylation of recombinant β-catenin in vitro and the fractions with the highest activities were tested for CKII inhibition. Heparin, a known CKII inhibitor decreased the amount of phosphorylated GST-β-catenin (not shown). To further proof the specifity of this inhibition, kinase activities were assayed with a competitor peptide containing CK II consensus motifs (Fig. 1A).

Phosphorylation of β-catenin was clearly reduced in the presence of CKII-specific competitor peptides (compareFig. 1A, lanes 3 and 4), but not with control peptides containing mutated CKII motifs(Fig. 1A, co). These results provide evidence that the heparin-mediated inhibition ofβ-catenin-phosphorylation is CKII-specific, and therefore CKII activity was measured in immunocomplexes collected with anti-β-catenin from cell lysates. The results indicated further that CKII and β-catenin are associated in vivo. Consistent with that, specific in vivo associations between the two proteins were found in 293- and NIH 3T3 cells in immunoprecipitations with anti-CKII-α antibodies and subsequent immunoblots with anti-β-catenin (Fig. 1B, lanes 2,4). Conversely, when Myc-tagged CKII-α was transiently expressed in 293 cells, CKII could also be co-precipitated with anti-β-catenin antibodies from cell lysates, whereas an unrelated kinase could not (Fig. 1B, lanes 7,8). This association might well be a direct interaction, since recombinant CKII is able to interact directly with GST-β-catenin in vitro(Fig. 1C, lane 5). This interaction becomes enhanced in the presence of ATP(Fig. 1C, lane 6), similar to previous observations for the association of CKII with other target proteins(Muslin et al., 1996;Pawson, 1995). Taken together,these results provide good evidence that CKII interacts with and phosphorylates β-catenin.

To determine which part of β-catenin is phosphorylated by CKII,various β-catenin deletion constructs were expressed and the purified GST fusion-proteins were subjected to in vitro kinase assays(Fig. 2A). Heparin was included in parallel as inhibitor of CKII phosphorylation. As shown inFig. 2B, strong phosphorylation was observed in the N-terminal (aa residues 1-119/1-302) and C-terminal (aa residues 683-781) regions of β-catenin. Remarkably, in contrast to the C-terminal region, the N-terminal phosphorylation was efficiently inhibited by heparin, indicating that CKII preferentially phosphorylates the N-terminal part of β-catenin. These results are consistent with the localization of three conserved CKII-consensus motifs in the N-terminal region ofβ-catenin around the GSK-3β recognition motifs(Fig. 2C).

In the following the phosphorylation of β-catenin by CKII was investigated with respect to its biological functions, the best studied of which are its central role in the E-cadherin-catenin adhesion complex and as a Wnt signaling component. In these experiments mutant forms of β-catenin were included where one or more of the three CKII consensus motifs were destroyed by single or combined aa substitutions (i.e. aa S29→A;T102→A; T112→Q) (Fig. 3A). GST fusion proteins with either single or combined mutations were tested in kinase assays in vitro. No phosphate-incorporation was observed when β-catenin mutated in all three CKII consensus motifs(Ser/Thr-mutant) was tested with recombinant CKII(Fig. 5D, lane 7), and single aa substitutions drastically reduced phosphorylation (not shown).

The cytoplasmic turn-over of wt and mutant β-catenin was compared in pulse-chase experiments with 293 cells stably expressing HA-tagged wt or Ser/Thr-mutant β-catenin (Fig. 3). At each time-point, HA-tagged β-catenin was immunoprecipitated from cell lysates containing comparable amounts of incorporated radioactivity and quantitated. The autoradiograph inFig. 3A shows that the time-dependent degradation of mutant β-catenin was clearly delayed as compared to the wt form. Three independent pulse-chase experiments were quantified using the phospho-imager (Fig. 3B). The half-life of HA-tagged wt β-catenin is similar to that previously determined for endogenous β-catenin(Aberle et al., 1997). In comparison, Ser/Thr-mutant β-catenin showed a significant stabilization by more than two-fold (Fig. 3B). From these results it is concluded that functional CKII-motifs are required for the control of the cytoplasmic amount ofβ-catenin. These results suggested further that CKII is part of the multi-protein complex which controls the cytoplasmic amount of β-catenin. To test this possibility Myc-tagged Axin, or Myc-GSK3β were transiently expressed in 293 cells, endogenous CKII was immunoprecipitated and the precipitates were probed with anti-Myc antibodies(Fig. 4A). Both, Myc-Axin(Fig. 4A, lane 2) and Myc-GSK3β (Fig. 4A, lane 4) were found associated with endogenous CKII suggesting that CKII is a component of the β-catenin degradation machinery. It is well established that phosphorylation of β-catenin by GSK3β enhances binding ofβ-catenin to Axin and APC. A similar notion can be taken from the results depicted in Fig. 4B, sinceβ-catenin mutated in its CKII-motifs can only poorly associate with Axin. In these experiments HA-tagged wt or Ser/Thr-mutant β-catenin and Myc-Axin were transiently expressed in 293 cells, comparable amounts of immunoprecipitates were collected with anti-Myc antibodies(Fig. 4B, lanes 4-6), and precipitates were probed for β-catenin(Fig. 4B, lanes 1-3). The introduction of one mutation in β-catenin already decreased the affinity to Axin (Fig. 4B, lane 2) and binding was further decreased when all three CKII-motifs were mutated(Fig. 4B, lane 3). These results demonstrate that β-catenin mutated in its CKII-motifs can poorly associate with Axin and suggested that both GSK3β and CKII act synergistically in controlling the degradation of β-catenin. To address this question, wt or Ser/Thr-mutant GST-β-catenin was phosphorylated with either CKII, GSK-3β, or both in vitro. The reaction was stopped with a high-salt wash (500 mM NaCl, 0.5% Triton X-100), and bound kinases were eluted and examined by immunoblot analysis with antibodies as indicated(Fig. 5). CKII bound to wtβ-catenin, but binding was clearly reduced with mutant β-catenin(compare Fig. 5A, lanes 3 and 7).

Importantly, enhanced binding of GSK-3β to β-catenin was observed when β-catenin was pre-phosphorylated by CKII(Fig. 5B, compare lanes 4 and 5). Pre-phosphorylation of Ser/Thr-mutant β-catenin with CKII did not enhance binding of GSK-3β to mutant β-catenin(Fig. 5B, lane 9) and binding of GSK-3β to wt and Ser/Thr-mutant β-catenin was comparable(Fig. 5B, lanes 4,8). InFig. 5D the[γ-³²P]ATP-incorporation in these experiments is depicted,demonstrating enhanced phosphate incorporation in wt β-catenin when both kinases were used (lane 5) and the lack of incorporated phosphate in the CKII mutant (lane 7). Comparable amounts of wt and Ser/Thr-mutant β-catenin were used in these experiments, as monitored with anti-GST antibodies(Fig. 5C).

Several conclusions can be drawn from these results. Most notably, CKII and GSK-3β bind and phosphorylate wt β-catenin synergistically and pre-phosphorylation by CKII enhances binding of GSK-3β. In the Ser/Thr-mutant CKII can still bind, although to a reduced amount, but no phosphate incorporation is observed here. Thus, the CKII consensus motifs inβ-catenin are substrates for the enzyme. Mutations in the CKII motifs do not affect binding and phosphorylation by GSK-3β, but the enhanced binding in the wt when both kinases were used is not observed for the mutant form. Altogether, a sequential action of CKII and GSK-3β in phosphorylation of β-catenin is suggested, and the role of CKII in controlling the protein turnover is underlined.

Since Ser/Thr-mutant β-catenin becomes stabilized, it was of interest to examine how the subcellular distribution of mutant β-catenin compares to wt protein. For this, 293 cells stably expressing HA-tagged wt or Ser/Thr-mutant β-catenin were subjected to sequential detergent extractions to separate the cytosolic, membrane-organelle and cytoskeletal fractions of cells. Immunoblot analysis with anti-HA antibodies revealed the expected subcellular distribution of transfected HA wt β-catenin, with the major content in the membrane-organelle fraction, similar to that of endogenous wt β-catenin in untransfected 293 cells(Fig. 6A). In comparison, the cytoplasmic pool of Ser/Thr-mutant β-catenin was drastically increased and little mutant protein was detected in the cytoskeletal fraction(Fig. 6A). In these preparations the HA antibody crossreact most likely with c-terminal degradation products from HA β-catenin, whereas a 65 kDa-band from the membrane-organelle fractions is of unrelated specifity. The reduced amount of Ser/Thr-mutant β-catenin in the cytoskeletal fraction was of particular interest and pointed to the possibility of an altered interaction of the mutant form with α-catenin.

Therefore, in vitro association experiments were carried out with recombinant α- and β-catenin(Fig. 6B,C). Wild-typeβ-catenin bound to increasing amounts of α-catenin. However, littleα-catenin interacted with Ser/Thr-mutant β-catenin(Fig. 6B). Interestingly,pre-phosphorylation of wt β-catenin with CKII in vitro even enhanced binding of α-catenin, but this was not the case for the mutant form ofβ-catenin (Fig. 6C). These results provide strong evidence that phosphorylation of β-catenin by CKII regulates the interaction between α- and β-catenin. They further explain the reduced amount of Ser/Thr-mutant β-catenin in the cytoskeletal fraction and its increase in the cytosol as seen inFig. 6A. This difference in sub-cellular localization could influence cell behaviour. To test this possibility, wound-healing experiments were performed with 293 cells stably expressing wt and Ser/Thr-mutant β-catenin(Fig. 7A). Remarkably, the wound closure was much faster in 293 cells expressing Ser/Thr-mutantβ-catenin compared to those expressing the wt form. The enhanced wound closure is very unlikely due to differences in cell proliferation between the two cell types, as monitored with the proliferation marker Ki-67(Fig. 7B). Instead, these results suggest that 293 cells expressing mutant β-catenin exhibit higher migratory potential.

This is also underlined by the subcellular distribution of HA-tagged wt or Ser/Thr-mutant β-catenin in these stably transfected cells(Fig. 7C). In comparison to HA wt β-catenin, which is strongly localized at cell-cell-contact sites,Ser/Thr-mutant β-catenin shows a more intense and diffuse staining in the cytoplasm. Since the switch of β-catenin from the membrane to the cytoplasm is required for cell-migration, this also reflects the higher migratory potential of the cells expressing Ser/Thr-mutant β-catenin.

A similar distribution can also be seen for α-catenin in both cell lines. However, significant amounts of α-catenin in mutant cells are still localized at the membrane due to the association with endogenousβ-catenin. In line with such a view, an increased staining for the mesenchymal marker vimentin in cells expressing Ser/Thr-mutant β-catenin could be observed (S.B., unpublished).

In our attempts to identify kinases which use β-catenin as substrate we immuno-precipitated β-catenin from cell lysates and tested the precipitates for kinase activities in vitro. We found that CKII associates with and phosphorylates β-catenin and also identified GSK-3β in theβ-catenin-associated precipitates (not shown). Even more interestingly,we detected a yet unknown kinase activity which specifically phosphorylates the C-terminal region (aa 683-731) of β-catenin(Fig. 2B) and which is distinct from CKII and GSK-3β as judged with specific inhibitors for both. It will be of future interest to identify and characterize this kinase. CKII phosphorylates preferentially the N-terminal region of β-catenin and this phosphorylation was specifically inhibited with a peptide containing CKII consensus motifs. We identified aa residues Ser29, Thr102, and Thr112 in the N-terminal region of β-catenin as target-sites for CKII and mutated these aa residues to compare this Ser/Thr-mutant and wt β-catenin in subsequent analysis.

It is well established that GSK-3β regulates the cytoplasmic turnover of β-catenin in a multiprotein complex and that the stability and composition of this complex is highly dynamic and depends on phosphorylation(Rubinfeld et al., 1996;Hart et al., 1998;Ikeda et al., 1998;Fagotto et al., 1999;Kawahara et al., 2000;Strovel et al., 2000). Here we propose that CKII participates together with GSK-3β in controlling the cytoplasmic stability of β-catenin and is part of the multiprotein complex. We show that CKII can associate with GSK-3β and Axin and that Ser/Thr-mutant β-catenin binds less efficiently to Axin and GSK-3β. Moreover, Ser/Thr-mutant β-catenin has a longer half-life(Fig. 3) as has been observed with β-catenin which is not phosphorylated by GSK-3β(Aberle et al., 1997). A synergistic phosphorylation by CKII and GSK-3β was observed when wt and Ser/Thr-mutant β-catenin were compared in in vitro kinase assays(Fig. 5). Pre-phosphorylation by CKII enhanced binding of GSK-3β and increased phosphorylation ofβ-catenin. The CKII-dependent higher binding and phosphorylation capabilities of GSK-3β are not observed in mutant β-catenin,suggesting a sequential action of CKII and GSK-3β in phosphorylatingβ-catenin. Taking this together with the biochemical data, it is concluded that CKII phosphorylation of β-catenin stabilizes its binding to Axin and GSK-3β and thus enhances the activity of GSK-3β. Our findings are in line with the demonstrated involvement of CKII in the proteolytic degradation of other proteins, e.g. IkBα and lens connexin 45.6 (Bren et al., 2000;Yin et al., 2000). Altogether it is concluded that a combined action of CKII and GSK-3β controls the cytoplasmic turnover of β-catenin.

Using the Wnt1-expressing mouse mammary epithelial cell line C57MG, a Wnt1-dependent increase of CKII expression resulting in stronger phosphorylation of β-catenin. Association of both proteins with Dvl proteins (the mammalian homologues of Drosophila Dishevelled) was previously observed, and it was suggested that CKII acts as a positive regulator of Wnt signaling (Song et al.,2000). Our results do not support such a view, but one cannot exclude that CKII activity on β-catenin at the membrane and in the cytoplasm vary in different cell types, which raises the question how this can be achieved by a widely distributed and nearly always active kinase.

Possible explanations come from GSK-3β, which is differentially regulated by insulin and Wnt signaling. As a consequence of different post-translational modifications (e.g. phosphorylation on serine-9) and complex formation (e.g. the APC-Axin-Frat/GBP-Dvl-complex), wnt-signals are able to accumulate β-catenin in the cytoplasm, whereas insulin-signals do not (Ding et al., 2000). Although no extracellular signal are confirmed to regulate CKII-activity,similar mechanisms can be discussed for CKII-mediated regulation ofβ-catenin. CKII-phosphorylation of β-catenin also might be regulated through the interaction with different protein-complexes in the respective subcellular compartments. In line with this, CKII-activity can be regulated through the interaction with p53 in the nucleus and with p47(phox)(Guerra et al., 1997;Kim et al., 2001).

Ser/Thr-mutant and wt β-catenin exhibit different sub-cellular distributions when stably expressed in 293 cells. The relative amount of Ser/Thr-mutant β-catenin in the cytosolic fraction is higher, consistent with the prolonged half-life of the mutant protein. Significantly, only little mutant β-catenin distributes to the cytoskeletal fraction, indicating that mutant β-catenin does not fulfil properly its role in the cadherin-catenin complex. Such a view is supported by our immunofluorescence data showing that Ser/Thr-mutant β-catenin is preferentially distributed in the cytoplasm whereas wt β-catenin is mostly localized to the cell membrane. This suggested that the affinity of Ser/Thr-mutant β-catenin to either E-cadherin and/or α-catenin might be reduced. To test this possibility protein interaction assays were performed with recombinant proteins. Binding of E-cadherin to wt or Ser/Thr-mutant β-catenin was unchanged (as was the binding of another β-catenin interaction partner,LEF-1; S.B., unpublished). However, mutant β-catenin bound less efficiently to α-catenin and, even more importantly, pre-phosphorylation of wt β-catenin with CKII enhanced its binding to α-catenin(Fig. 6B,C). These results provide strong evidence that CKII regulates binding of β-catenin toα-catenin and thus add a new molecular mechanism to modulate the E-cadherin-catenin cell adhesion complex. In 293-cells the catenins are propably sequestered to another cadherin, since these cells do not express E-cadherin. This view is supported by immunoprecipitation-experiments in ³⁵S-labeled 293-cells. After separation of the anti β-catenin immuno-complex, we could detect three bands running at the size of E-cadherin,α- and β-catenin, as could be observed in cells containing E-cadherin (S.B., unpublished). Again, the biochemical data are underlined by the distribution of α-catenin, which shows significant similarity with the staining of HA-tagged wt or Ser/Thr-mutant β-catenin in immunofluorescence analysis. However, significant amounts of α-catenin in mutant cells is still localized at the membrane due to the association with endogenous β-catenin.

From the N-terminal aa-sequence of β-catenin, a amphipathicα-helix can be predicted, which has been confirmed for the aa 134-161(Graham et al., 2000). However for formation of the β-catenin-α-catenin heterodimer a change in the overall secondary structure of β-catenin is necessary(Huber and Weis, 2001). The introduction of negatively charged phosphate-residues in neighbouring sequences from β-catenin might help to provide the surface for this hydrophobic interactions. Whereas Tyr 142 in β-catenin is absolutely needed for the interaction with the hydrophobic core of the binding region(Aberle et al., 1996b;Huber et al., 1997;Pokutta and Weis, 2000), the residues targeted by CKII might regulate the strength of the interaction.

Related to the interaction with α-catenin, a hydrophobic binding pocket within the armadillo repeats 3 and 4 in β-catenin, is critical for Axin-binding (Graham et al.,2000; von Kries et al.,2000). N-terminal phosphorylation of β-catenin by CKII might provide the structure of the hydrophobic interaction surface. Binding domains for E-cadherin, APC and LEF-1/Tcf in β-catenin are not or are only partially located in this region. Consequently, the binding to these proteins was not influenced after CKII phosphorylation of β-catenin in vitro(S.B., unpublished).

Our biochemical results provide convincing evidence that phosphorylation ofβ-catenin by CKII modulates the biological function of β-catenin. The wound-healing experiments additionally substantiate this. Cells expressing Ser/Thr-mutant β-catenin exhibit higher migratory potential as compared to cells expressing the wt protein. An increase in cell proliferation is unlikely to account for the enhanced wound healing; instead, cells expressing mutant β-catenin appear to have adopted a more mesenchymal phenotype. The enhanced migratory potential of cells expressing Ser/Thr-mutant β-catenin is likely due to the inefficient binding of mutant protein to α-catenin,which could affect the cytoskeletal architecture of the cells. Although other explanations are possible, this view is supported by the subcellular distribution of HA-tagged wt or Ser/Thr-mutant β-catenin andα-catenin. Mutations in aa residues in β-catenin which are targets for CKII could thus be relevant in tumorigenesis. Notably, aa residue Ser29,which we found to be phosphorylated by CKII, is mutated in some gastric-cancer cell lines, and results in an accumulation of β-catenin(Polakis, 2000;Park et al., 1999).

We are grateful to Li Zeng, Columbia University, New York for providing us with the Myc-tagged Axin-1 plasmid, and to P. E. Shaw for the Myc-tagged ERK2 plasmid. We thank Rosemary Schneider for secretarial work and Randy Cassada for critical reading of the manuscript.