Regulation of β-Catenin Function by the IκB Kinases

Both the β-catenin and the nuclear factor κB (NF-κB) proteins are important regulators of gene expression and cellular proliferation. Two kinases, IKKα and IKKβ, are critical activators of the NF-κB pathway. Here we present evidence that these kinases are also important in the regulation of β-catenin function. IKKα- and IKKβ-deficient mouse embryo fibroblasts exhibited different patterns of β-catenin cellular localization. IKKβ decreases β-catenin-dependent transcriptional activation, while IKKα increases β-catenin-dependent transcriptional activity. IKKα and IKKβ interact with and phosphorylate β-catenin using both in vitro and in vivo assays. Our results suggest that differential interactions of β-catenin with IKKα and IKKβ may in part be responsible for regulating β-catenin protein levels and cellular localization and integrating signaling events between the NF-κB and Wingless pathways.

␤-Catenin, the mammalian homologue of the Drosophila armadillo protein, is a ubiquitously expressed protein that has at least two distinct roles in the cell. First, it participates in cell-cell adhesion by mediating the association of E-cadherin with the cytoskeleton (1,2). Second, it is a critical downstream component of the Wnt 1 or Wingless signal transduction pathway (3)(4)(5). The Wnt family of secretory glycoproteins plays an important role in embryonic development, in the induction of cell polarity, and in the determination of cell fate. Deregulation of Wnt signaling disrupts axis formation in embryos (5)(6)(7)(8) and is associated with multiple human malignancies (9).
The current model of Wnt signaling indicates that the binding of the Wnt proteins to their receptor, frizzled, stabilizes ␤-catenin by inhibiting the activity of a serine/threonine kinase glycogen synthase kinase-3 or GSK-3␤ (9). GSK-3␤ is associated with ␤-catenin in a multiprotein complex that includes the adenomatous polyposis coli tumor suppressor protein (APC), axin or conductin, protein phosphatase 2A, and dishevelled. GSK-3␤ phosphorylation of residues in the amino terminus of ␤-catenin results in APC-mediated ␤-catenin degradation via the ubiquitin-proteosome pathway (10,11). Increased levels of ␤-catenin are frequently found in colon cancer due to mutations in either the APC gene (12)(13)(14) or at residues in the amino terminus of ␤-catenin that are phosphorylated by GSK-3␤ (15)(16)(17). In the nucleus, ␤-catenin forms a complex with members of the T-cell factor (TCF)/lymphocyte-enhancer factor (LEF) family and activates gene expression of a variety of target genes (18 -23) including c-myc (24) and cyclin D1 (25,26).
NF-B comprises a family of transcription factors which are critical in activating the expression of genes involved in the immune and inflammatory response and in the regulation of cellular apoptosis (27,28). NF-B is sequestered in the cytoplasm by a family of inhibitory proteins known as IB. Upon stimulation of this pathway by a variety of agents including IL-1 and TNF␣, the kinases IKK␣ and IKK␤ (29 -33) in conjunction with the scaffold protein IKK␥/NEMO (34 -36) leads to the phosphorylation of IB␣ at serine residues 32 and 36. Gene disruption studies in mice indicate that IKK␤ appears to be the critical kinase in activating the NF-B pathway (37)(38)(39), while IKK␣ appears to be critical for other functions such as keratinocyte differentiation (40 -42). IKK␣ and IKK␤ can form homodimers and also heterodimerize with each other, and this process is critical for their kinase activity. IKK phosphorylation of IB␣ leads to its ubiquitination and degradation by the 26S proteasome and the nuclear translocation of NF-B (43).
Interestingly, the sequence DSGXXS, which is the target site in IB for IKK phosphorylation, is also found in the amino terminus of ␤-catenin (11). Phosphorylation of this sequence in both ␤-catenin and IB stimulates their interactions with the ubiquitin ligase ␤-TrCP leading to their degradation by the proteasome (10,11,44). It has also been demonstrated that the ␤-catenin/TCF complex increases ␤-TrCP levels by a posttranscriptional mechanism to result in opposite effects on ␤-catenin and NF-B activity (45). In addition, disruption of either the murine GSK-3␤ and IKK␤ genes result in a similar phenotype with embryonic lethality due to hepatic apoptosis from increased sensitivity to TNF␣ (46). These results suggest potential relationships between ␤-catenin and NF-B signaling pathways.
Given the fact that both the NF-B and ␤-catenin pathways are important in the control of cellular proliferation and are regulated by cellular kinases that lead to ␤-TrCP-mediated degradation (10,11,45), we explored potential similarities in their regulation. First, we addressed whether there were differences in the cellular localization of ␤-catenin in wild-type mouse embryo fibroblasts as compared with fibroblasts derived from IKK␣-and IKK␤-deficient mice. Next, we analyzed inter-actions between both IKK␣ and IKK␤ and ␤-catenin and determined whether these kinases regulated ␤-catenin-dependent transcriptional activity. The results of this analysis indicate that IKK␣ can positively regulate ␤-catenin-dependent transcriptional activity while IKK␤ negatively regulates this activity.

MATERIALS AND METHODS
Cells and Reagents-SW480 cells were purchased from American Tissue Culture Collection (Manassas, VA) and maintained in L-15 medium supplemented with 10% fetal bovine serum (HyClone Laboratories), 2 mM L-glutamine, and antibiotics (penicillin-streptomycin). COS, mouse embryo fibroblasts (MEFs, a kind gift of Xiaodong Wong), IKK␣ and IKK␤ knock-out cells (39,42) were maintained in Dulbecco's modified Eagle's medium and supplemented with the same components as above.
Plasmid Constructs-The pCMV5 expression plasmids containing either FLAG-tagged IKK␣ and IKK␤ including the constitutively active kinases (SS/EE) with substitutions at residues 176/180 for IKK␣ or 177/181 for IKK␤ and the kinase defective (K/M) mutants at residue 44 in both IKK␣ and IKK␤ were described previously (30,47,48). Wildtype and mutant IKK␣ and IKK␤ cDNAs were each cloned into the baculovirus expression vector pAcHLT. The recombinant baculoviruses were used to infect SF9 cells to produce recombinant IKK proteins for in vitro kinase assays (48). The pCMV5 expression vectors containing the wild-type and the dominant negative NIK mutant in which lysine residues at positions 429 and 430 were substituted with alanine contained an amino-terminal Myc-tag (48).
The pCMV5 expression vectors encoding full-length human ␤-catenin was provided by S. Byers, while the plasmids for LEF-1, TOP-FLASH, and FOPFLASH were gifts of K. Kinzler and R. Grosschedl. The RSV-␤-galactosidase construct was a gift from P. Chaudhary. The glutathione S-transferase (GST) full-length ␤-catenin (GST ␤-cat-(1-781)) bacterial expression vector was constructed by using polymerase chain reaction (PCR) to generate a fragment encompassing the fulllength ␤-catenin, which was then cloned in frame with GST in the pGEX. The GST fusion protein containing the amino-terminal 91 amino acids of ␤-catenin was constructed by SacI digestion and ligation of the GST fusion containing wild-type ␤-catenin. The constructs GST-␤-cat-(130 -781), GST-␤-cat-(1-400), GST-␤-cat-(130 -400), and GST-␤-cat-(618 -781) were constructed using PCR. The amino-terminal deleted form of ␤-catenin utilized PCR primers to generate a fragment containing amino acids 130 -781, which was cloned into pCMV5 and contained a carboxyl-terminal HA-epitope. All constructs that were generated by PCR were subjected to DNA sequencing and cloned into pcDNA3.
Expression and Purification of GST ␤-Catenin Fusion Proteins-Recombinant GST ␤-catenin fusion proteins were expressed in bacterial strain BL21 and lysed in HMK buffer (50 mM Tris (pH 7.5), 0.1 M NaCl, 1 mM phenylmethylsulfonyl fluoride), and the bacterial lysates were incubated with 0.5 ml of packed glutathione-conjugated-agarose beads (Sigma) for 2 h at 4°C. After three washes, the GST fusion proteins were eluted with 10 mM glutathione in HMK buffer and dialyzed, and protein purity was assessed by SDS-polyacrylamide gel electrophoresis.
Luciferase Reporter Assays-COS cells and mouse embryo fibroblasts were plated at 50% confluence in 35-mm tissue culture wells. After 24 h, the cells were transfected using LipofectAMINE Plus with the indicated DNA constructs and either the TOPFLASH luciferase reporter containing LEF/TCF binding sites or the FOPFLASH luciferase reporter with mutated LEF/TCF sites. An NF-B luciferase reporter containing three NF-B binding sites upstream of a thymidine kinase minimal promoter was used to detect NF-B-directed gene expression. An RSV-␤-galactosidase expression vector was included in the transfection assays to control for differences in transfection efficiency, and the pCMV5 plasmid was added to the transfection assays to standardize DNA quantities. Between 18 to 24 h posttransfection, the cells were washed twice with cold PBS, and the reporter activity was measured using the luciferase assay system (Promega). All transfections were done in duplicate and repeated at least three times.
Protein Association and Western Blot Analysis-For endogenous protein association studies, equal volumes of proteins (200 -300 l) from each of the Superdex-200 fraction were incubated overnight at 4°C with 1 g of indicated antibodies or normal mouse IgG followed by the addition of protein G-agarose (Sigma) for 2-3 h at 4°C. For protein association studies using transfected IKK and ␤-catenin expression vectors, COS cells were transfected with FLAG-tagged IKK␣ or IKK␤ and HA-tagged ␤-catenin cDNAs. Cells were harvested 18 -24 h after transfection, extracts were prepared, and gel chromatography was performed as described above. Equal volumes of each column fraction were immunoprecipitated with 12CA5 antibody or anti-FLAG M2 antibody. Immunoprecipitates were resolved by SDS-polyacrylamide gel electrophoresis, transferred onto nitrocellulose membranes (Amersham Pharmacia Biotech), and probed with specific antibodies. The membranebound immune complexes were analyzed with an enhanced chemiluminescence system (Amersham Pharmacia Biotech). For in vitro association studies, 40 l of the cytoplasmic fractions were incubated overnight at 4°C with 40 l of the glutathione-conjugated-agarose bound with indicated proteins. Following three washes with 10 volumes of cold PBS, the protein complexes were subjected to Western blot analysis as described above.
In Vitro Kinase Assays-Kinase assays were performed as described by Yamamoto et al. (48). The baculovirus-produced IKK proteins were purified by nickel-agarose chromatography and then immunoprecipitated with 12CA5 monoclonal antibody (48). The epitope-tagged IKK␣ and IKK␤ kinases were transfected into COS cells, and extracts were immunoprecipitated with the M2 monoclonal antibody directed against the FLAG-epitope. These kinases were added to kinase buffer containing 10 Ci of [␥-32 P]ATP, 1 mM ATP, 1 mM dithiothreitol, 5 mM MgCl 2 , 100 mM NaCl, 50 mM Tris-HCl (pH 8.0), and then 1 g of each of the substrates including wild-type or the S32A/S36A of GST-IB␣-(1-54) or GST-␤-cat-(1-91) were incubated for 15 min at 30°C. For determination of phosphate incorporation into the GST-IB␣-(1-54) and GST-␤catenin-(1-91), 2 g of each of these substrates was incubated with either FLAG-tagged IKK␣ or IKK␤, which were immunoprecipitated from COS cell extract with the M2 monoclonal antibody in kinase buffer containing 15 Ci of [␥-32 P]ATP with a specific activity of 6000 Ci/mM (New England Nuclear) and either 0.01 mM, 0.01 mM, or 1.0 mM of cold ATP. The kinase reaction mixtures were subjected to electrophoresis on 10% SDS-polyacrylamide gels and autoradiography. The 32 P-labeled IB␣ and ␤-catenin substrates were then subjected to scintillation counting, and the moles of phosphate incorporated were calculated. Reactions were incubated at 30°C for 5, 15, 30, 60, and 120 min and stopped by the addition of protein loading buffer and heating to 90°C.
Immunocytochemistry and Confocal Microscopy-Cells were cultured overnight on coverslips in Dulbecco's modified Eagle's medium without serum, washed two times with PBS, and fixed with 3.7% formaldehyde for 10 min followed by a brief permeabilization with 0.5% Triton X-100 in PBS. The cells were blocked for 30 min with 3% normal donkey serum in PBS and then incubated for 1 h with primary antibodies (diluted 1:50 to 1:200 in 1% normal donkey serum in PBS). The coverslips were washed three times with PBS and then incubated for 1 h with the secondary antibodies conjugated with FITC or Red-X rhodamine (diluted 1:400 in 1% normal donkey serum in PBS). Samples were washed three times and then treated with Aquamount (Polysciences). The results were analyzed on a laser scanning confocal microscope MRC 1000 (Bio-Rad).
Immunostaining of MEFs, IKK␣-, and IKK␤-deficient cells with a polyclonal antibody directed against ␤-catenin demonstrated that ␤-catenin has a different pattern of staining in IKK␣ Ϫ/Ϫ and IKK␤ Ϫ/Ϫ cells. ␤-Catenin was present in both the nucleus and the cytoplasm of MEFs with marked accumulation at cell-cell junctions (Fig. 1B, panel A). In IKK␣ Ϫ/Ϫ cells, there was reduced nuclear staining of ␤-catenin as compared with MEF cells (Fig. 1B, panel G). There was more ␤-catenin present in the nucleus and the perinuclear region of IKK␤ Ϫ/Ϫ cells than in IKK␣ Ϫ/Ϫ cells (Fig. 1B, panel G). As a control, these cells were also stained with a monoclonal antibody directed against the basal transcription factor TFIIB, which is localized predominantly in the nucleus (Fig. 1B, panels B, E, and H). There was little background staining when the FITC-conjugated anti-goat secondary antibody was used with the mouse monoclonal antibody directed against TFIIB (Fig. 1B, panels C, F, and I). These results indicate that there is less ␤-catenin localized in the nucleus of IKK␣ Ϫ/Ϫ cells than in either IKK␤ Ϫ/Ϫ cells or MEF cells.
␤-Catenin Activity in IKK-deficient Cells-Next we addressed whether the differences in ␤-catenin distribution in the IKK-deficient embryo fibroblasts could alter its transcriptional activity. The IKK-deficient cells and the parental MEFs were transfected with a TOPFLASH reporter construct alone or with expression vectors encoding either LEF-1 or ␤-catenin. The TOPFLASH reporter is driven by four LEF/TCF binding motifs inserted upstream of a minimal c-fos promoter and a luciferase gene (15). As a control, the FOPFLASH reporter, which lacks LEF/TCF binding sites, was utilized. An RSV-␤-galactosidase expression vector was included in these transfections to control for differences in transfection efficiency.
When the TOPFLASH reporter alone was transfected into IKK␣ Ϫ/Ϫ cells, there was consistently a 5-6-fold lower level of activity as compared with that observed in IKK␤ Ϫ/Ϫ cells (Fig.  2). Transfection of an expression vector encoding LEF-1 into either IKK␣ Ϫ/Ϫ or IKK␤ Ϫ/Ϫ cells markedly stimulated TOP-FLASH activity as did transfection of expression vectors encoding both LEF-1 and ␤-catenin. There was no significant activity from the FOPFLASH reporter in either the absence or presence of ␤-catenin and LEF-1 (Fig. 2). Transfection of expression vectors encoding wild-type IKK␤ and LEF-1 into IKK␣ Ϫ/Ϫ cells reduced TOPFLASH activity, while transfection of an IKK␣ expression vector with LEF-1 increased TOPFLASH activity in these cells (Fig. 2). When similar studies were performed in IKK␤ Ϫ/Ϫ cells, transfection of an IKK␤ expression vector reduced TOPFLASH activity, while transfection of an IKK␣ ex-

FIG. 1. Characterization of ␤-catenin localization in IKK␣-and IKK␤-deficient cells.
A, MEF (panels A and B) and either IKK␣-deficient (IKK␣ Ϫ/Ϫ ) (panels C and D) or IKK␤-deficient (IKK␤ Ϫ/Ϫ ) (panels E and F) embryo fibroblasts were plated overnight on coverslips before staining with either rabbit polyclonal antibodies directed against IKK␣ (panels A, C, and E) or IKK␤ (panels B, D, and F) followed by staining with a secondary Red-X rhodamine-conjugated rabbit antibody. B, alternatively MEF (panels A-C), IKK␣ Ϫ/Ϫ (panels D-F) and IKK␤ Ϫ/Ϫ (panels G-I) cells were stained with either a goat antibody to ␤-catenin (green) (panels A, D, and G) or a mouse monoclonal antibody to TFIIB (red) (panels B, E, and H). Donkey anti-goat antibody conjugated with FITC and anti-mouse conjugated with Red-X rhodamine were used as secondary antibodies. The respective negative controls utilizing the TFIIB monoclonal antibody and the donkey anti-goat FITC-conjugated antibody are also shown (panels C, F, and I). Images were collected using a laser scanning confocal microscope (Bio-Rad).
pression vector did not significantly alter TOPFLASH activity (Fig. 2). Transfection of both ␤-catenin and LEF-1 resulted in similar levels of TOPFLASH activity in the IKK␤ Ϫ/Ϫ and IKK␣ Ϫ/Ϫ cells (Fig. 2). The parental MEF cells consistently gave somewhat higher TOPFLASH activity than seen in the IKK-deficient cells (Fig. 2). Again transfection of an IKK␤ expression vector with LEF-1 into these cells reduced TOP-FLASH activity, while transfection of an IKK␣ expression vector with LEF-1 resulted in little change in TOPFLASH activity (Fig. 2). Thus, the reduced levels of endogenous ␤-catenin in the nuclei of IKK␣ Ϫ/Ϫ cells are associated with decreased ␤-catenin activation of gene expression, and this defect could be complemented by transfection of an IKK␣ expression vector. IKK␣ does not increase gene expression in the IKK␤ Ϫ/Ϫ and MEF cells, which have relatively abundant levels of nuclear ␤-catenin.
IKK␤ and IKK␣ Have Differential Effects on ␤-Catenin Transactivation-The results presented in the previous section suggested that IKK␤ and IKK␣ could potentially be involved in regulating the transcriptional stimulatory properties of ␤-catenin. Thus it was important to address whether either IKK␣ or IKK␤ could alter ␤-catenin-mediated transcriptional activation in COS cells, which have low levels of endogenous ␤-catenin in the nucleus and relatively low levels of IKK␣ and IKK␤ (data not shown). COS cells were transfected with either a TOP-FLASH or FOPFLASH reporter, LEF-1 and ␤-catenin expression vectors, and increasing amounts of expression vectors encoding either the wild-type, constitutively active or kinasedefective mutants of IKK␣ and IKK␤. The constitutively active IKK proteins (IKK␣ SS/EE and IKK␤ SS/EE) have glutamate substituted for serine residues in their T-loop so as to mimic phosphorylation of these residues and increase the activity of these kinases (30). The kinase-defective mutants (IKK␣ K/M and IKK␤ K/M) contain a substitution of a lysine residue at position 44 with methionine (30).
As previously demonstrated, the coexpression of ␤-catenin and LEF-1 increased TOPFLASH but not FOPFLASH activity (Fig. 3A). When either wild-type IKK␤ or the constitutively activate kinase, IKK␤ SS/EE, was cotransfected with ␤-catenin and LEF-1, TOPFLASH activity decreased in a concentration-dependent manner (Fig. 3A). In contrast, cotransfection of either wild-type IKK␣ or the constitutively active kinase, IKK␣ SS/EE, increased ␤-catenin-dependent transactivation in a concentration-dependent manner (Fig.  3A). Transfection of the IKK␤ K/M mutant resulted in a modest decrease in ␤-catenin transactivation that was not concentration-dependent, while transfection of the IKK␣ K/M mutant did not significantly alter ␤-catenin transactivation (Fig. 3A).
The cotransfection experiments in COS cells indicated that IKK␣ increased ␤-catenin-dependent gene expression, while IKK␤ decreased ␤-catenin-dependent gene expression. Thus, we investigated whether IKK␣ and IKK␤ can alter ␤-catenin protein levels. In addition, we asked whether IKK␣ and IKK␤ would affect the protein levels of a ␤-catenin mutant in which serine residues 33 and 37 were changed to alanine to result in increased ␤-catenin protein levels (15)(16)(17). Expression vectors encoding either the hemagglutinin-tagged wild-type or S33A/ S37A mutant ␤-catenin were transfected into COS cells either alone or in the presence of either the constitutively active FLAG-tagged IKK␣ or IKK␤. Whole cell extracts were prepared from the transfected COS cells and analyzed by Western blot analysis using the 12CA5 and M2 monoclonal antibodies directed against the hemagglutinin and FLAG epitopes, respectively (Fig. 3B). IKK␣ expression increased the level of the epitope-tagged ␤-catenin protein (Fig. 3B, lane 2), while IKK␤ reduced the amount of the epitope-tagged ␤-catenin protein (Fig. 3B, lane 3). In contrast, IKK␣ did not alter the level of the S33A/S37A ␤-catenin mutant (Fig. 3B, lane 5), while IKK␤ reduced the level of this protein (Fig. 3B, lane 6). Transfection assays with the TOPFLASH reporter indicated that IKK␣ increased gene expression in the presence of the wild-type but not the mutant ␤-catenin, while IKK␤ reduced gene expression in the presence of both of these ␤-catenin proteins (data not shown). These results suggest that IKK␣ either directly or indirectly may lead to increased levels of ␤-catenin to increase TOPFLASH activity, while IKK␤ may reduce the levels of ␤-catenin to decrease ␤-catenin activity. The failure of IKK␣ to further increase the protein levels of the mutant ␤-catenin suggests that the structure of the amino terminus of ␤-catenin may be important in this process.
It was important to determine whether activation of the NF-B pathway may be involved in the increased TOPFLASH activity seen in the presence of ␤-catenin and LEF-1. The TOPFLASH reporter was transfected with expression vectors encoding ␤-catenin, LEF-1, and the IB␣ super-repressor (IB␣ SS/AA) (Fig. 3C). The IB␣ super-repressor protein, which contains substitutions of serine residues 32 and 36 with ala-nine, cannot be phosphorylated by IKK, and its resistance to degradation prevents the nuclear translocation of the NF-B proteins in response to activators of this pathway (27). The transfection of the IB␣ super-repressor did not alter activation of the TOPFLASH reporter in the presence of ␤-catenin and LEF-1 expression vectors, while it completely abolished the activity of an NF-B reporter (Fig. 3C). These results suggest that NF-B activation does not appear to be involved in the activation of TOPFLASH activity by ␤-catenin and LEF-1.
The Amino Terminus of ␤-Catenin Is Critical for IKK␣ but Not IKK␤ Modulation of Gene Expression-Next we addressed whether the same or different domains in ␤-catenin were required for regulation by IKK␣ and IKK␤. The amino terminus of ␤-catenin is phosphorylated by GSK-3␤ leading to ␤-catenin degradation (50 -52). Amino-terminal deletion mutants of ␤-catenin are very stable because they lack sequences that are involved in APC-mediated degradation (50 -54). Furthermore, our results suggested that the amino terminus of ␤-catenin may be involved in IKK␣-mediated regulation. To determine whether the amino terminus of ␤-catenin was critical for mediating the effects of IKK␣ and IKK␤, transfection of increasing amounts of a ␤-catenin expression vector deleted of its first 129 amino acids was transfected into COS cells along with LEF-1. There was increased TOPFLASH activity seen with this mutant similar to the results seen with wild-type ␤-catenin (Fig. 4). The expression of the constitutively active IKK␤ protein reduced activation of TOP-FLASH reporter when transfected with this ␤-catenin mutant. In contrast, the expression of the constitutively active IKK␣ protein did not alter the ability of the amino-terminal deletion of ␤-catenin to activate the TOPFLASH reporter (Fig. 4). These results suggest that the decreased ␤-catenin transactivation observed with IKK␤ is not dependent on the amino terminus of ␤-catenin, while IKK␣ requires the presence of this domain to stimulate ␤-catenin transcriptional activity.  6). Whole cell extracts were prepared and subjected to Western blot analysis with the anti-HA monoclonal antibody 12CA5 to detect the HA-tagged ␤-catenin (lanes 1-6, top panel) or the M2 monoclonal antibody directed against the FLAG-epitope to detect FLAG-tagged IKK␣ and IKK␤ (lanes 1-6, lower panel). C, the TOPFLASH reporter was cotransfected with expression vectors encoding LEF-1 and ␤-catenin and the IB␣ super-repressor (0.1, 0.2 and 0.5 g) vectors. In addition, the NF-B luciferase reporter (100 ng) was transfected with expression vectors encoding either the constitutively active IKK␤ SS/EE or IKK␣ SS/EE and the IB␣ super-repressor. All COS cell transfections contained 0.5 g of an RSV-␤-galactosidase expression vector, and the DNA amounts were standardized with a pCMV5 vector. Extracts were prepared 18 h after transfection, and the normalized luciferase activity was determined by correcting for differences in ␤-galactosidase activity. The change in gene expression relative to the TOPFLASH reporter alone was determined for each transfection and the average of three experiments (each in duplicate) is shown.

␤-Catenin
Interacts with IKK␣ and IKK␤-To address whether the effects of IKK␣ and IKK␤ on ␤-catenin-dependent gene expression may be mediated by direct interactions with ␤-catenin, we performed coimmunoprecipitation experiments of ␤-catenin and the IKK proteins using cytoplasmic extracts prepared from the SW480 colon cancer cell line. SW480 cells express a truncated APC gene product and result in enhanced levels of ␤-catenin. This increased level of ␤-catenin was necessary to detect this protein in Western blot analysis of column fractions that were generated following chromatography (14).
Superdex-200 gel filtration chromatography of the S100 cytoplasmic extract prepared from SW480 cells was utilized to assay interactions between the IKK and ␤-catenin. Similar chromatographic analysis has previously been used to characterize the high molecular weight IKK complex (55). Western blot analysis of these column fractions indicated that ␤-catenin was present in a broad peak, including a portion that was present in high molecular weight fractions that also contained IKK␣ and IKK␤ (Fig. 5A, left panel). Column fractions 7-12, which contained both ␤-catenin and the IKK proteins, were immunoprecipitated with a monoclonal antibody directed against ␤-catenin followed by Western blot analysis with either ␤-catenin, IKK␣, or IKK␤ antibodies (Fig. 5A, middle panel). This analysis indicated that ␤-catenin was associated with IKK␣ and IKK␤, while immunoprecipitation of these column fractions with mouse IgG followed by Western blot analysis demonstrated no association of these proteins (Fig. 5A, right  panel). These results suggest that endogenous ␤-catenin can associate with IKK␣ and IKK␤.
Next, we characterized the interactions of IKK␣ and IKK␤ with ␤-catenin following cotransfection of COS cells with expression vectors encoding these epitope-tagged proteins (Fig.  5B). First, S100 extracts prepared from these cells were subjected to Superdex-200 chromatography and Western blot analysis. As previously noted, when IKK␣ and IKK␤ were transfected into COS cells they migrate in a relatively broad peak following Superdex-200 chromatography due to the failure to completely assemble into the high molecular weight IKK complex (49) (Fig. 5B, left and middle panels). Next, immunoprecipitation of column fractions prepared from extracts containing the FLAG-tagged IKK␣ and HA-tagged ␤-catenin was performed followed by Western blotting. This analysis indicated that FLAG-tagged IKK␣ and HA-tagged ␤-catenin were able to associate (Fig. 5B, left panel). Column fractions of extracts prepared from COS cells cotransfected with FLAGtagged IKK␤ and HA-tagged ␤-catenin indicated that both of these proteins were also able to associate (Fig. 5B, middle  panel). Western blot analysis of the immunoprecipitated IKK and ␤-catenin proteins do not strictly overlap. This is likely due to the fact that their elution profiles following chromatography vary, which is consistent with the presence of these proteins in multiple complexes. The column fractions containing HAtagged ␤-catenin and either FLAG-tagged IKK␣ or IKK␤ were also immunoprecipitated with mouse IgG and analyzed by Western blot analysis. This analysis revealed that there were not nonspecific associations of the ␤-catenin and IKK proteins (Fig. 5B, right panel).
To further characterize the interactions of ␤-catenin with IKK␣ and IKK␤, in vitro binding of SW480 cytoplasmic extract with GST proteins fused to different domains of ␤-catenin was performed. Thus, we could determine the role of different domains of ␤-catenin including the amino terminus, which regulates protein stability, the armadillo repeats, and the C-terminal transactivation domain in binding the IKK proteins (53). Following the incubation of the SW480 cytoplasmic extract with the GST-␤-catenin fusion proteins bound to glutathione-Sepharose beads, Western analysis was performed with antibodies directed against either IKK␣ or IKK␤. Each of the ␤-catenin fusion proteins, but not GST alone, was able to interact with IKK␣ and IKK␤ (Fig. 6B). However, the GST-␤catenin fusion proteins extending between amino acid residues 1-400 and 130 -400 consistently bound more IKK␣ and IKK␤ (Fig. 6B). These results suggested that the region of ␤-catenin containing the first six armadillo repeats was probably critical for interaction with the IKK proteins. The data from the GSTpull down assays in conjunction with coimmunoprecipitation data of both endogenous and transfected proteins demonstrate that the IKK proteins and ␤-catenin can interact under a variety of different conditions. IKK␣ and IKK␤ Phosphorylate ␤-Catenin-Next we addressed whether IKK could phosphorylate the amino terminus of ␤-catenin and whether stimulation of IKK activity could result in increased ␤-catenin phosphorylation in in vitro kinase assays. The amino terminus of ␤-catenin has been demonstrated to be a target for GSK-3␤ phosphorylation (9), while serine residues 32 and 36 in the amino terminus of GST-IB␣ are the target for IKK phosphorylation (29 -33). HeLa cells were either untreated, treated with TNF␣, or transfected with an expression vector encoding NIK (56, 57) to induce IKK kinase activity. The IKK complex was immunoprecipitated from extracts prepared from these cells and assayed for its FIG. 4. Differential effects of IKK␤ and IKK␣ on transactivation of an amino-terminal truncated ␤-catenin. COS cells were cotransfected with the indicated plasmids including the TOPFLASH reporter, LEF-1, and either wild-type ␤-catenin (group 1) or an aminoterminal deletion of the first 129 amino acids of ␤-catenin (groups 2-4). The construct encoding the amino-terminal-deleted ␤-catenin was cotransfected at concentrations of 0.5 (group 2), 1.0 (group 3), and 2.0 g (group 4) together with the constitutively active kinases IKK␤ SS/EE (0.5 and 1.0 g) and IKK␣ SS/EE (0.5 and 1.0 g). An RSV-␤-galactosidase expression vector was added to each transfection, and DNA quantities were standardized by addition of a pCMV5 expression vector. After 18 h, the cells were collected and luciferase activity was determined and normalized to correct for differences in ␤-galactosidase activity. The change in gene expression relative to the TOPFLASH reporter alone was determined for each transfection and the average of three experiments (each in duplicate) is presented. ability to phosphorylate either GST-␤-cat-(1-91), GST-IB␣-(1-54), or GST-IB␣ (SS/AA)-(1-54). IKK activity was induced by treatment with either TNF␣ or NIK and increased the phosphorylation of ␤-catenin (Fig. 7A, lanes 1-3) and IB␣ (Fig.  7A, lanes 4 -6), but not the IB␣ mutant in which serine residues 32 and 36 were changed to alanine (Fig. 7A, lanes 7-9).
Recombinant baculovirus-produced IKK␣ and IKK␤ were also tested in in vitro kinase assays using GST fusions with ␤-catenin or IB␣. Both IKK␣ and IKK␤ also phosphorylated the amino terminus of ␤-catenin and IB␣, but not the IB␣ mutant (Fig. 7B). COS cells were next transfected with either epitope-tagged wild-type or mutant IKK␣ and IKK␤, and following immunoprecipitation with the M2 monoclonal antibody these kinases were assayed using in vitro kinase assays with ␤-catenin and IB␣ as substrates (48). Wild-type IKK␣ and IKK␤, but not the kinase-defective mutants, were able to phosphorylate ␤-catenin and IB␣ (Fig. 7C).
Finally, we addressed whether IKK␣ and IKK␤ could also phosphorylate additional regions in ␤-catenin other than its amino terminus (Fig. 7D). Both kinases phosphorylated GST fusion proteins containing various portions of ␤-catenin (Fig.  7D, lanes 2-5). These GST fusions contained either the amino terminus of ␤-catenin, an amino-terminal-deleted form of ␤-catenin or full-length ␤-catenin (Fig. 7D). Similar results were obtained using IKK␣ and IKK␤ preparations produced by baculovirus expression (data not shown). These results indicate that both IKK␣ and IKK␤ phosphorylate multiple regions of ␤-catenin.
Stoichiometry of IKK Phosphorylation of IB␣ and ␤-Catenin-Next we compared the ability of IKK␣ and IKK␤ to phosphorylate GST-IB␣-(1-54) and GST-␤-cat-(1-91) substrates. In these in vitro kinase assays, we analyzed the phosphorylation of each of these substrates at specific points over a 120-min time course utilizing 0.01 mM, 0.1 mM, and 1.0 mM of cold ATP and 15 Ci of [␥-32 ]ATP. Following SDS-PAGE and autoradiography (Figs. 8A and 8B, top panels), the 32 P-incorporation into the ␤-catenin and IB␣ substrates was determined, and the number of the moles of phosphate incorporated per mole of  [7][8][9][10][11][12], which contained the ␤-catenin and the IKK proteins, were incubated overnight with monoclonal antibodies directed against ␤-catenin (Transduction Laboratories). Western blot analysis was then performed on these immunoprecipitates using polyclonal goat antibody to detect ␤-catenin or rabbit polyclonal antibodies to detect IKK␣ and IKK␤ (middle panel). These column fractions were also immunoprecipitated with mouse IgG followed by Western blot analysis with polyclonal antibodies directed against either ␤-catenin, IKK␣, or IKK␤ (right panel). B, COS cells were transfected using LipofectAMINE Plus with expression vectors encoding HA-tagged ␤-catenin and either FLAG-tagged IKK␣ or FLAG-tagged IKK␤. Cells were harvested 18 h after transfection, and S100 extracts were prepared and fractionated on a Superdex-200 column. Equal column volumes (40 l) were subjected to Western blot analysis using 12CA5 antibody to detect HA-tagged ␤-catenin (left and middle panels) or the M2 monoclonal antibody to detect FLAG-tagged IKK␣ (left panel) or FLAG-tagged IKK␤ (middle panel). Immunoprecipitation of column fractions prepared from FLAG-tagged IKK␣ and HA-tagged ␤-catenin (left panel) or FLAG-tagged IKK␤ and HA-tagged ␤-catenin (middle panel) transfected cells was performed using either the 12CA5 or M2 monoclonal antibodies followed by Western blot analysis with the antibody that was not used in the immunoprecipitation. Column fractions from extracts containing either FLAG-tagged IKK␣ and HA-tagged ␤-catenin (right panel, first and third gels) or FLAG-tagged IKK␤ and HA-tagged ␤-catenin (right panel, second and fourth gels) were immnoprecipitated with mouse IgG followed by Western blot analysis with either 12CA5 (HA) or M2 (FLAG) antibodies as indicated.
substrate was calculated (Fig. 8, A and B, lower panels).
Kinase assays performed with IKK␣ using 1 mM of cold ATP resulted in ϳ0.05 mol of phosphate/mol of protein incorporated into the amino terminus of ␤-catenin as compared with 0.09 mol of phosphate/mol of protein incorporated into the amino terminus of IB␣ after a 120-min reaction (Fig. 8A). Kinase assays performed with IKK␤ using 1 mM of cold ATP demonstrated that there was 1.4 mol of phosphate/mol of protein incorporated into the amino terminus of ␤-catenin and 0.5 mol of phosphate/mol of protein incorporated in the amino terminus of IB␣ after 120 min (Fig. 8B). It is interesting to note that the phosphorylation of the ␤-catenin by IKK␤ may be biphasic in contrast to its phosphorylation of IB␣ (Fig. 8B). Similar phosphate incorporation into these substrates was found using both baculovirus-produced and COS-transfected IKK␣ and IKK␤ proteins (data not shown). In agreement with previous studies, this analysis indicates that IKK␣ is a much weaker kinase than is IKK␤ in phosphorylating IB␣ (58) and ␤-catenin. These results indicate that the IKK proteins result in relatively similar incorporation of phosphate into the amino terminus of ␤-catenin and IB␣, although there are differences in the kinetics of this process.

DISCUSSION
In this study, we present data that IKK␣ and IKK␤ can modulate ␤-catenin function. First, we observed the differential localization of ␤-catenin in mouse embryo fibroblasts derived from IKK␣-and IKK␤-deficient cells. Second, the transcriptional activity of ␤-catenin was higher in IKK␤ Ϫ/Ϫ cells as compared with IKK␣ Ϫ/Ϫ cells. Third, IKK␤ decreased ␤-catenindependent gene expression similar to the effects seen with GSK-3␤, while IKK␣ increased this activity. Fourth, we found that IKK␣ expression in COS cells increased the amount of ␤-catenin, while IKK␤ expression reduced the amount of ␤-catenin. Finally, we demonstrated that IKK␣ and IKK␤ interacted with and were able to phosphorylate ␤-catenin. Experiments are underway to map the sites in ␤-catenin that are phosphorylated by IKK␣ and IKK␤ in order to determine whether phosphorylation alters ␤-catenin function. Our preliminary results suggest that IKK␣ phosphorylates different residues in the amino terminus of ␤-catenin than serine residues 33 and 37 that are phosphorylated by GSK-3␤.
Studies with an amino-terminal deletion of ␤-catenin indicated that IKK␣ requires this region to increase ␤-catenin-de-FIG. 6. In vitro interaction of ␤-catenin with endogenous IKK␣ and IKK␤. A, a schematic representation of the GST ␤-catenin fusion proteins that were used to analyze interactions with SW480 extract is shown. B, GST-fusion proteins with ␤-catenin were bound to glutathione-Sepharose beads (lanes 1-6), and after overnight incubation with SW480 cell extract, the glutathione beads were extensively washed and Western blot analysis was performed using rabbit polyclonal antibodies directed against either IKK␣ or IKK␤ as indicated. 10% of the SW480 lysate alone is shown in lane 7. C, the GST-fusion proteins were analyzed by Coomassie Blue staining. pendent gene expression, while the effects of IKK␤ on ␤-catenin activity are not dependent on this region. These results and the finding that IKK␣ is not able to increase the protein levels of an amino-terminal ␤-catenin mutant suggest that the amino terminus of ␤-catenin is likely involved in IKK␣ regulation. Thus, IKK␣ and IKK␤ likely have effects on different domains of ␤-catenin to alter its role on gene expression. It is unclear whether differences in the kinase activity of IKK␣ and IKK␤ are involved in their differential effects on ␤-catenin-dependent gene expression or whether other effects such as differential binding to specific pools of ␤-catenin may be involved. Finally, additional mechanisms such as IKK effects on ␤-catenin protein stability and/or nuclear import or export are possible. Multiple factors including Wnt signaling, the TCF/LEF proteins (19,52,53), and APC (59) affect the cellular localization of ␤-catenin, which lacks a canonical nuclear localization signal. Although we demonstrate that the IKK proteins interact with ␤-catenin, it is possible that IKK interaction with other com-ponents of the Wnt pathway such as APC may also be involved in regulating ␤-catenin function.
Both IKK␣ and IKK␤ can form heterodimers and homodimers, and dimerization of these kinases is essential for their activity (33,47,60,61). However, previous data has suggested that there is no synergy between IKK␣ and IKK␤ in regulating their kinase activity (58). Given the wide disparity in their kinase activity, they may have other cellular targets in addition to IB (58). The ability of these kinases to potentially associate with as yet unidentified cellular factors may alter their substrate specificity. Gene disruption studies indicate that IKK␤ rather than IKK␣ is the critical kinase involved in the activation of the NF-B pathway in response to treatment with either TNF␣ or IL-1␤ (37)(38)(39). The predominant cytoplasmic localization of IKK␤ probably reflects the major role of this kinase in the phosphorylation of the IB proteins that are localized in the cytoplasm bound to the RelA/p65 NF-B protein (37)(38)(39). The results of our immunofluorescence studies FIG. 7. IKK␣ and IKK␤ phosphorylate ␤-catenin and IB␣. A, HeLa cells were either untreated, treated with TNF␣ (20 ng/ml) for 10 min, or transfected with a pCMV5 expression vector encoding NIK. Extracts were immunoprecipitated with a polyclonal antibody directed against IKK␣ and IKK␤ and in vitro kinase assays were performed with GST fusions with ␤-catenin-(1-91), IB␣-(1-54) or IB␣ SS/AA-(1-54) followed by autoradiography. B, recombinant baculovirus-expressed IKK␣ and IKK␤ proteins were purified as described. In vitro kinase assays were performed using IKK␣ and IKK␤ and GST-␤-cat-(1-91) (lanes 1 and 4) and GST-IB␣-(1-54) (lanes 2 and 5) as substrates (upper panel). A GST-IB␣ protein in which serine residues 32 and 36 were changed to alanine (lanes 3 and 6) was also assayed. The GST fusion proteins used in the in vitro kinase assay were monitored by Coomassie Blue staining (lower panel). C, COS cells were transfected with expression vectors encoding either FLAG-tagged wild-type or mutant IKK␣ or IKK␤ kinases or mock transfected (no kinase). After 30 h, the cells were collected, and cellular extract was immunoprecipitated with the M2 monoclonal antibody directed against FLAG-epitope. The upper panel demonstrates in vitro phosphorylation of either GST-␤-catenin (lanes 1, 4, 7, 10, and 13), GST-IB␣ (lanes 2, 5, 8, 11, and 14) or no added substrate (lanes 3, 6, 9, 12, and 15) by the indicated kinases. Expression of the transfected IKK constructs was analyzed by Western blot analysis (middle panel). Immunoprecipitation of extracts from mock-transfected cells were also analyzed in in vitro kinase assays (lanes [13][14][15]. The amount of GST-␤-catenin and GST-IB␣ substrates used in these assays was monitored by Coomassie  A and B, top panels). Incorporation of 32 P into these substrates was quantitated by scintillation counting and the moles of phosphate incorporated per mole of substrate was calculated (A and B, bottom panels).
suggest that IKK␣ is localized in both the nucleus and cytoplasm of MEFs and may be predominantly nuclear in the absence of IKK␤ in IKK␤ Ϫ/Ϫ cells. Consistent with these observations, Western blot analysis of extracts prepared from COS cells transfected with expression vectors encoding IKK␣ and IKK␤ indicate that IKK␤ is predominantly localized in the cytoplasm, while IKK␣ is present in both the nucleus and the cytoplasm. Additional studies are currently underway to better characterize the cellular localization of IKK␣. Whether any of the effects of IKK␣ on skin and skeletal development may in part be mediated by either IKK␣ binding and/or phosphorylation of ␤-catenin remains to be determined.
IKK regulation of ␤-catenin activity differs from its activation of the NF-B pathway. Cytokines such as TNF␣ stimulate IKK phosphorylation of IB leading to its rapid degradation and the nuclear translocation of NF-B. TNF␣ activation of an NF-B reporter construct is blocked by transfection of an IKK␤ dominant negative mutant (30). Although TNF␣ treatment of cells results in marked decreases in ␤-catenin-dependent gene expression, this effect is only partially blocked by an IKK␤ dominant negative mutant. These results suggest that the effects of TNF␣ on ␤-catenin-dependent gene expression likely involve additional substrates and/or pathways other than IKK␤ and ␤-catenin. Although our results support a role for IKK␣ and IKK␤ on modulating ␤-catenin activity, the regulation of this pathway is different from that seen with TNF␣induction of IKK to activate the NF-B pathway.
Several observations are also consistent with the potential for similar factors being involved in the regulation of the Wnt and NF-B pathways. It has been demonstrated that ␤-catenin/ TCF signaling increases ␤-TrCP levels by a posttranscriptional mechanism to result in increased degradation of both ␤-catenin and IB (45). Thus, changes of ␤-TrCP levels can result in marked effects on both the ␤-catenin and NF-B pathways. Additionally, GSK-3␤, which is an important kinase involved in regulating ␤-catenin levels, has also been implicated in regulating NF-␤ activation. Gene disruption studies have indicated that GSK-3␤ Ϫ/Ϫ mice have a phenotype similar to IKK␤deficient mice developing liver degeneration as a result of increased sensitivity to TNF␣ stimulation (46). The mechanism by which GSK-3␤ may alter the NF-B pathway remains to be determined. In summary, our studies suggest that a common set of cellular factors may be involved in the integration of a variety of cellular signaling processes that regulate the NF-B and ␤-catenin pathways.