Interaction of FOXO with β-Catenin Inhibits β-Catenin/T Cell Factor Activity*

Wingless (Wnt) signaling regulates many aspects of development and tissue homeostasis, and aberrant Wnt signaling can lead to cancer. Upon a Wnt signal β-catenin degradation is halted and consequently the level of β-catenin in the cytoplasm increases. This allows entry of β-catenin into the nucleus where it can regulate gene transcription by direct binding to members of the lymphoid enhancer factor/T cell factor (TCF) family of transcription factors. Recently, we identified Forkhead box-O (FOXO) transcription factors as novel interaction partners of β-catenin (Essers, M. A., de Vries-Smits, L. M., Barker, N., Polderman, P. E., Burgering, B. M., and Korswagen, H. C. (2005) Science 308, 1181-1184). Here we show that the β-catenin binding to FOXO serves a dual effect. β-catenin, through binding, enhances FOXO transcriptional activity. In addition, FOXO competes with TCF for interaction with β-catenin, thereby inhibiting TCF transcriptional activity. Reduced binding between TCF and β-catenin is observed after FOXO overexpression and cellular oxidative stress, which simultaneously increases binding between β-catenin and FOXO. Furthermore, small interfering RNA-mediated knock down of FOXO reverts loss of β-catenin binding to TCF after cellular oxidative stress. Taken together, these results provide evidence for a cross-talk mechanism between FOXO and TCF signaling in which β-catenin plays a central regulatory role.

Wnt proteins are closely related secreted glycoproteins that act as growth factors on cells and play critical roles in cell proliferation and cell fate determination at many stages of development (2,3). Genetic and biochemical experiments in Drosophila melanogaster, Xenopus laevis, and mammalian cells have established a framework for the Wnt signaling pathway. In the absence of a Wnt signal, cytoplasmic ␤-catenin is bound to a multi-protein ␤-catenin destruction complex that contains several proteins, including Axin, adenomatous polyposis coli (APC), 3 casein kinase I␣ (CKI␣) and ⑀ (CKI⑀) and glycogen synthase kinase 3 (GSK-3). In this complex, CKI␣ and/or CKI⑀ and GSK phosphorylate ␤-catenin (4 -6). Phosphorylation triggers ubiquitination of ␤-catenin by ␤TrCP, a component of the SCF ␤TrCP ubiquitin-protein ligase complex and degradation of ␤-catenin by the ubiquitin-proteasome pathway (7)(8)(9). In the presence of Wnt, Dishevelled blocks ␤-catenin degradation by inducing the disassembly of the ␤-catenin destruction complex, thereby allowing accumulation of ␤-catenin within the cytosol and entry into the nucleus (10). Within the nucleus, ␤-catenin can bind to lymphoid enhancer factor/T cell factor (TCF) family of transcription factors and induce transcription of Wnt target genes (11)(12)(13)(14)(15). Deregulation of the Wnt signaling pathway, for example due to loss of APC, results in stabilization and nuclear accumulation of ␤-catenin and results in tumor formation (16).
The FOXO subfamily of transcription factors is critically involved in the regulation of apoptosis, proliferation, and the control of oxidative stress (reviewed in Ref. 17). FOXOs are negatively regulated by the phosphoinositide-3 kinase/protein kinase B pathway. Activation of phosphoinositide-3 kinase/ protein kinase B will induce phosphorylation and nuclear exclusion of FOXO, thereby inhibiting FOXO transcriptional activity. Recently, we and others have obtained evidence that FOXOs are also controlled by oxidative stress. In contrast to insulin signaling, increased cellular oxidative stress relocalizes FOXO to the nucleus and results in FOXO activation (18). Activation by increased cellular oxidative stress requires phosphorylation by JNK and this is evolutionary conserved. In D. melanogaster and Caenorhabditis elegans dFOXO and DAF-16 are also phosphorylated by JNK, and JNK activity increases lifespan in D. melanogaster and C. elegans in a dFOXO/DAF1-16 dependent manner (19,20). Recently we showed that ␤-catenin directly binds to FOXO and that this binding leads to enhanced FOXO transcriptional activity (1). The binding of ␤-catenin to FOXO is increased under conditions of oxidative stress, and genetic analysis in C. elegans demonstrated that the interaction between FOXO and ␤-catenin is conserved and thus reveals an evolutionary conserved function for ␤-catenin, independent of TCF. Consistent with regulation of FOXO by phosphoinositide-3 kinase and Ras signaling, ligand-independent activation of FOXO causes a cell cycle arrest in G 1 , both in cells trans-formed with oncogenic Ras and cells with a deletion of, or a mutation in, phosphatase and tensin homolog deleted on chromosome 10 (PTEN) (21). Interestingly, colon carcinoma cells transformed by activated ␤-catenin/TCF signaling due to an APC truncation are also arrested in G 1 by activation of FOXO (22), suggesting that FOXO expression can suppress ␤-catenin/ TCF signaling toward proliferation. Therefore, we here investigated the consequences of the binding between ␤-catenin and FOXO for signaling through the ␤-catenin-TCF complex. We show that activation of FOXO leads to inhibition of TCF transcriptional activity. This suggested that interaction of FOXO and ␤-catenin competes with the binding of ␤-catenin with TCF. Consistent with this, increased cellular oxidative stress resulting from H 2 O 2 treatment reduced the TCF/␤-catenin interaction and this is likely mediated via FOXO as siRNAmediated knockdown of FOXO expression alleviated H 2 O 2 -induced reduction in ␤-catenin-TCF complex formation. Taken together, these data provide evidence for a novel mechanism of cross-talk between Wnt and phosphoinositide-3 kinase signaling whereby ␤-catenin acts as a pivot between essential downstream elements of these signaling pathways, TCF and FOXO, respectively.

EXPERIMENTAL PROCEDURES
Cell Culture, Transfection, and Infection-DL23 cells, DLD1 cells expressing a conditionally active FOXO3a.A3-ER fusion, were created as described (22). DLD1 human colon carcinoma cells, DL23 cells, and LS174T human colon carcinoma cells were maintained in RPMI 1640 supplemented with L-glutamine, penicillin/streptomycin, and 10% fetal calf serum. DL23 cells were treated with 500 nM 4-hydroxy-tamoxifen (4OHT) for 8, 16, or 24 h to activate the fusion protein. Human embryonic kidney 293T cells and Phoenix cells expressing the amphotrophe receptor were maintained in Dulbecco's modified Eagle's medium supplemented with L-glutamine, penicillin/ streptomycin, and 10% fetal calf serum. Human embryonic kidney 293T, DLD1, and DL23 cells were transiently transfected using FuGENE 6 reagent according to the manufacturer (Roche Applied Science). Total amounts of transfected DNA were equalized using pBluescript KSIIϩ.
LS174T cells were infected with pBabe-puro or pBabe-FOXO3a.A3 virus. For virus production Phoenix cells were transfected using SuperFect transfection reagent (3 mg/ml) with pBabe-puro or pBabe-FOXO3a.A3. Two days after transfection medium of the transfected Phoenix cells was harvested. LS174T cells were infected by adding the medium of the Phoenix cells together with 6 g/ml hexadimethrine bromide. Infection was repeated 6 h after the first round. The day after infection, cells were seeded in selection medium containing 2 g/ml puromycin.
Antibodies-Monoclonal 12CA5 and 9E10 antibodies were produced using hybridoma cell lines. Monoclonal antibody ␤-catenin was obtained from Transduction Laboratories. Monoclonal FLAG antibody was obtained from Sigma. Polyclonal HA antibody was obtained from Santa Cruz Biotechnology. The ABC antibody recognizing dephosphorylated ␤-catenin was generously provided by Mascha van Noort and Hans Clevers (Hubrecht Laboratory, The Netherlands).
Immunoprecipitation and Western Blots-Non-confluent cells were lysed in radioimmune precipitation buffer (50 mM Tris, pH 7.5, 1% Triton X-100, 0.5% deoxycholate, 10 mM EDTA, 150 mM NaCl, 50 mM NaF, 1 mM sodium vanadate, 1 g/ml leupeptin, and 0.1 g/ml aprotinin), and lysates were cleared for 10 min at 14,000 rpm at 4°C. Lysates were incubated for 2 h at 4°C with either 1 l of 12CA5 or 9E10 antibody or 7.5 l of ␤-catenin antibody and 50 l of pre-washed protein-A beads. The immunoprecipitations were washed four times with radioimmune precipitation buffer and cleared for all liquid, and 25 l of 1ϫ Laemmli sample buffer was added. Samples were subjected to SDS-PAGE and transferred to polyvinylidene difluoride membrane (PerkinElmer). Western blot analysis was performed under standard conditions and using the indicated antibodies.
Luciferase Reporter Assays-DLD1 and DL23 cells were transiently transfected with either a reporter construct bearing multiple copies of an optimal TCF-binding site (pTOPglow) or a reporter construct bearing multiple copies of a mutant form of the optimal TCF-binding site (pFOPglow) (25). DLD1 cells were cotransfected with HA-FOXO4, HA-FOXO4⌬DB, HA-HOXO3a, or a control plasmid. DL23 and DLD1 cells were treated with 500 nM 4OHT for the indicated times. Luciferase counts were normalized using Tk-Renilla-luciferase. Cells were washed twice with phosphate-buffered saline and lysed in passive lysis buffer, and luciferase activity was analyzed using a luminometer and a dual-luciferase assay kit according to the manufacturer (Promega). The -fold induction of luciferase activity on the TOPglow construct was divided by the -fold induction on the Fopglow construct, and this TCF Optimal Promoter (TOP)/Fake Optimal Promoter (FOP) ratio was plotted.

FOXO Inhibits ␤-Catenin/TCF-dependent Transcription-
To investigate whether the interaction between ␤-catenin and FOXO would affect the function of ␤-catenin as a co-activator of TCF transcriptional activity, we first analyzed the effect of FOXO activation on ␤-catenin/TCF-dependent transcription. To this end we used DL23 cells, which are DLD1 colon carcinoma cells expressing a conditionally active HA-FOXO3a. A3-ER fusion protein (26). FOXO3a activity can be switched on in these cells by treating cells with the estrogen analog 4OHT (26). Because of a mutation in the APC gene, these cells contain high levels of stabilized ␤-catenin and thus display high ␤-catenin/TCF transcriptional activity. Activation of FOXO3a in these cells, induced by 4OHT treatment, resulted in inhibition of TCF-dependent transcription as measured by the TOP/FOP reporter assay (25) (Fig. 1A). Activation of the TOPglow reporter, which contains multiple copies of an optimal TCFbinding site, was greatly reduced by activated FOXO, whereas FOXO had no effect on the FOPglow reporter, which contains multiple copies of a mutant form of the TCF-binding site (25). Also, transient expression of FOXO3a and FOXO4 in DLD1 cells strongly inhibited TCF transcriptional activity, suggesting this effect not to be specific to FOXO3a or the inducible FOXO3a-A3-ER fusion protein (Fig. 1B). Thus, activation of FOXOs inhibits the activity of the TCF transcription factor.
Inhibition of TCF transcriptional activity following FOXO activation may occur through various mechanisms. FOXO activation may result in ␤-catenin degradation. However, total ␤-catenin levels were not affected by 4OHT treatment (Fig. 1A). In addition, we also analyzed whether the observed inhibition of TOP/FOP reporter activity was due to reduced levels of ␤-catenin not phosphorylated on its GSK-3 sites ("active" ␤-catenin). To determine this pool of ␤-catenin not phosphorylated at its GSK-3 sites, we performed immunoblotting using the ABC antibody (Fig. 1A, lower panel). Again, similar to total ␤-catenin, we did not observe a major change in dephosphorylated ␤-catenin. Recently, the use of antibodies detecting unphosphorylated ␤-catenin levels has become a subject of discussion (27), and it may be argued that the specificity of this type of antibodies is not sufficient. Irrespective of this discussion, it can be concluded that the observed inhibition of TOP/ FOP reporter activity is not likely the result of increased GSK-3-mediated ␤-catenin phosphorylation. The observed inhibition of TOP/FOP reporter activity following FOXO activation could also result from FOXO competing with TCF for binding to the TCF DNA-binding sites present in the TOPglow reporter. We analyzed in vitro binding of FOXO to known TCF DNA binding elements. Whereas in vitro the DNA binding domain of FOXO4 fused to glutathione S-transferase (GST-FOXO4-DB) did bind to the IRE (T(G/A)TTT motif-containing insulin-response element) present within the IGFBP-1 promoter, it failed to interact with the TCF-binding sites present within the Ultrabithorax homeotic gene promoter (14) (data not shown). This indicates that competition for DNA binding is unlikely. Furthermore, in vivo a FOXO4 mutant lacking the DNA binding domain (FOXO4⌬DB) was still able to inhibit ␤-catenin/TCF transcriptional activity (Fig. 1B). Thus, inhibition of TCF signaling by FOXOs is independent of FOXO DNA binding activity and therefore not due to direct competition between FOXO and TCF for binding to the TCF consensus DNA sequence.
Activation of FOXOs in DLD1 cells causes a cell cycle arrest in G 1 (26). To exclude that the inhibition of TCF signaling by FOXOs was not secondary to the ability of FOXOs to induce a cell cycle arrest, we performed the TOP/FOP reporter assay in the presence of the cell cycle inhibitors p16 INK4 or p21 WAF1 . Expression of these cell cycle inhibitors in DL23 and DLD1 cells induced a G 1 arrest (data not shown), and this did result in a partial reduction of TCF-dependent transcription (Fig. 1C), suggesting that indeed a cell cycle arrest in G 1 in these cells may contribute to the reduced ␤-catenin/TCF activity. More importantly, TCF activity was still further reduced upon activation of FOXO3a in the DL23 cells (Fig. 1C), indicating that although a cell cycle arrest may contribute to the inhibition of TCF signaling by FOXO this inhibition cannot account fully for the effect of FOXO. Thus, FOXO-mediated inhibition is not just secondary to the ability of FOXOs to induce a cell cycle arrest in these cells.
Previously, we showed that ␤-catenin could bind both FOXO3a and FOXO4 (1). Similar to FOXO3a, overexpression of FOXO4 inhibited TCF transcriptional activity, indicating that the ability of FOXOs to inhibit TCF signaling correlates with their ability to bind ␤-catenin. Thus, from these experiments we conclude that the observed inhibition of TOP/FOP reporter activity following FOXO activation most likely results from ␤-catenin binding to FOXO and consequently sequestration of ␤-catenin away from TCF.
FOXO Reduces Binding of ␤-Catenin to TCF-Previously, we have shown that FOXO binds to armadillo repeat 1-7 of ␤-catenin, the same region to which TCF binds (1, 11), suggesting that competition between FOXO and TCF for ␤-catenin indeed may occur. To establish more directly whether interaction between FOXO and ␤-catenin could compete with the interaction between TCF and ␤-catenin, we next analyzed binding between TCF and ␤-catenin in DL23 cells. Activation of FOXO3a in DL23 cells by adding 4OHT reduced the interaction between ␤-catenin and TCF4 ( Fig. 2A). 4OHT treatment of the control DLD1 cells did not affect this interaction. Furthermore, time course analysis showed that decreased ␤-catenin/TCF interaction could already be observed after 8 h of 4OHT treatment (Fig. 2B). This is consistent with the time course of inhibition of TOP/FOP reporter activity as shown in Fig. 1A. Next we analyzed interaction of HA-TCF4 and ␤-catenin in 293T cells. Following stabilization of ␤-catenin through LiCl-mediated GSK-3 inhibition, substantial binding of HA-TCF4 and ␤-catenin is observed. Co-expression of FOXO4 resulted in reduced HA-TCF4/␤-catenin interaction (Fig. 2C). Taken together these data suggest that FOXO and TCF-4 compete for interaction with ␤-catenin and that the inhibition of TCF signaling by FOXO reflects the ability of FOXO to compete with TCF for the interaction with ␤-catenin. Although the above provided evidence that FOXO can compete with TCF for ␤-catenin binding, the reverse may also occur. To test whether TCF can reduce the association between FOXO and ␤-catenin we analyzed oxidative stress-induced binding between FOXO4 and ␤-catenin in the absence or presence of increasing levels of exogenously expressed TCF (Fig. 2D). Increasing levels of TCF resulted in loss of stress-induced binding between FOXO4 and ␤-catenin, indicating competition occurs in a reciprocal manner.
Peroxide Stress-induced TCF/␤-Catenin Inhibition Is Mediated via FOXO-In DL23 cells FOXO activation occurs after 4OHT treatment. Treatment of cells with oxidative stress also results in activation of FOXO (18). Furthermore, following treatment of cells with increasing amounts of oxidative stress the interaction between FOXO and ␤-catenin is induced and transcriptional activity of FOXO is enhanced (1). Thus it is predicted that cellular oxidative stress will impact on TCF tran-scriptional activity. Indeed, increasing concentrations of H 2 O 2 resulted in a dose-dependent decrease in TCF transcriptional activity (Fig. 3A) and a dose-dependent decrease in HA-TCF4 interaction with ␤-catenin (Fig. 3B). To establish whether the observed decrease in HA-TCF4 binding to ␤-catenin after H 2 O 2 treatment of cells is due to increased interaction of ␤-catenin with FOXO following H 2 O 2 treatment of cells, we made use of small interfering RNA-mediated knockdown of FOXO4 in 293T cells. H 2 O 2 -induced reduction of HA-TCF4-␤-catenin complex formation was largely rescued by small interfering RNA-mediated knockdown of FOXO4 (Fig. 3C). As 293T cells express both FOXO3a and FOXO4 (28) this seems to suggest that FOXO4 is the main FOXO interacting with ␤-catenin. These results are consistent with a model that FOXO and TCF compete for ␤-catenin interaction and that cellular oxidative stress regulates this by diverting ␤-catenin from TCF to FOXO.
Inhibition of Endogenous ␤-Catenin/TCF Target Genes by Activation of FOXO-Finally, we tested whether expression of endogenous ␤-catenin/TCF target genes can be inhibited by activation of FOXOs. To this end we compared DL23 and DLD1 cells induced with 4OHT (Fig. 4A) or LS174T colon carcinoma cells infected with either a control virus or a constitutively active FOXO construct, FOXO3a.A3 (Fig. 4B). Quantitative PCRs for several target genes were performed on the isolated RNA from these cells. As described, FOXO3a activation in DL23 cells induced the expression of the known FOXO target gene p130 (Fig. 4A) (26). On the other hand, expression of cyclin D1 was inhibited (Fig. 4A). Cyclin D1 has been described as a target gene for the ␤-catenin-TCF complex (29), and inhibition of cyclin D1 expression by FOXO has been described previously (30,31). Importantly, inhibition of cyclin D1 expression by FOXO occurs at the transcriptional level (30,31), although the cyclin D1 promoter does not contain a bona fide FOXO DNA binding element (31) and inhibition was shown to be independent of binding of FOXO to the cyclin D1 promoter region (30). The cyclin D1 inhibition can be explained by the competition of FOXO and TCF to interact with ␤-catenin. However, gene expression regulation often summarizes the action of multiple signaling pathways, and thus cyclin D1 regulation does not necessarily reflect the status of only TCF and FOXO signaling or the interaction between TCF and FOXO. Therefore, to exclude as much as possible interference of other signaling pathways we also analyzed other TCF target genes. Activation of FOXO3a by 4OHT treatment in DL23 cells also repressed the mRNA expression of the TCF target genes Pitx2 (32) and Ephrin B2 (6) (Fig. 4A). This result demonstrated that activation of FOXO3a inhibits the expression of multiple TCF target genes. To exclude that the repression of TCF target genes is specific to the DL23 cells and/or the 4OHT induction system, we also analyzed the effect of expression of active FOXO3a in LS174T colon carcinoma cells. Both DLD1 and LS174T cells are colon carcinoma cells with elevated levels of ␤-catenin, although the differentiation status of both cell lines is different. LS174T cells harbor mutated ␤-catenin with wild type APC and p53, whereas DLD1 cells have wild type ␤-catenin but mutant APC and p53. Again, we observed repression of Ephrin B2 mRNA levels by FOXO3a (Fig. 4B). Also, the mRNA expression level of tetraspanin-5, a ␤-catenin/TCF-dependent target gene specific to LS174T as compared with DLD-1 cells, 4 was repressed by FOXO3a (Fig. 4B). Thus, FOXO3a can repress the transcriptional activation of multiple ␤-catenin/TCFdependent target genes in cell types that differ with respect to the mechanism by which ␤-catenin is stabilized, and therefore the effect of FOXO3a on endogenous gene regulation is again likely to be at the level of ␤-catenin. Taken together, these data provide evidence that the interaction between ␤-catenin and FOXO inhibits endogenous TCF-dependent gene transcription.

DISCUSSION
The data presented here suggest a model in which ␤-catenin provides a link between the WNT signaling pathway and the 4 N. Barker, personal communication. oxidative stress/FOXO pathway. We show that FOXO inhibits TCF-dependent transcription by binding to ␤-catenin and therefore functionally competes with TCF. Peroxide stress can strengthen the interaction between FOXO and ␤-catenin and potently inhibit TCF-dependent transcription. Furthermore, we show that FOXO is important for the peroxide stress-induced inhibition of TCF/␤-catenin interaction. Competition between FOXO and TCF is reciprocal as we also observed a dose-dependent decrease in stress-induced FOXO/␤-catenin interaction following TCF overexpression. Furthermore, TCF and FOXO bind to the same part of ␤-catenin (1) and we cannot detect the presence of FOXO in TCF immunoprecipitates. 5 Thus, we conclude that FOXO and TCF compete for the same pool of active ␤-catenin. Consequently, FOXO can inhibit TCF and vice versa TCF can inhibit FOXO signaling, and therefore they act through ␤-catenin as interdependent negative regulators. It is becoming apparent that the role of ␤-catenin in regulating gene transcription is not restricted to TCF. Besides binding and regulating FOXO, binding of ␤-catenin to HIF-1␣ and c-jun has been reported (33,34). Kaidi et al. (33) show that ␤-catenin can interact with HIF1␣ in a way similar to FOXO. During hypoxia the ␤-catenin/TCF interaction is inhibited whereas the ␤-catenin/HIF1␣ interaction is enhanced. This leads to higher expression of HIF1␣ target genes whereby ␤-catenin promotes adaptation to hypoxia. Once hypoxia is being resolved ␤-catenin binds again to TCF and proliferation of the cells can be restored. ␤-Catenin and c-jun also interact, albeit indirectly via TCF. This creates in a ␤-catenin-mediated manner a positive feedback loop increasing the expression of c-jun. The interaction is phosphorylation (JNK)-dependent and thus integrates both the JNK and APC/␤-catenin pathways (34).
In addition to regulation of ␤-catenin-mediated TCF transcriptional activity through the destruction complex, Wnt signaling also regulates so-called non-canonical Wnt signaling that involves stress-activated kinases like JNK and Nemo-like kinase. Interesting from the perspective of the negative regulation of TCF activity following oxidative stress as we observed here, non-canonical signaling also inhibits TCF activity. Thus, Wnt signaling harbors a built-in control of dampening TCF activity based upon stress signaling cascades. In contrast to TCF, JNK through direct phosphorylation of FOXO increases FOXO activity (18), suggesting JNK controls a switch from TCF inactivation to FOXO activation under conditions of oxidative stress. However, in JNK Ϫ/Ϫ cells FOXO expression still reduces TCF transcriptional activity, indicating that JNK is at least not essential. 5 A recent study by Funato et al. (35) reported that oxidative stress results in activation rather than inhibition of TCF transcriptional activity. By interacting with Dishevelled, nucleoredoxin inhibits Wnt/␤-catenin signaling. Peroxide stress decreases the interaction between nucleoredoxin and Dishevelled leading to the stabilization of ␤-catenin and subsequent activation of TCF. These results are in contrast to the observations reported here. A possible explanation for this difference could reside within the kinetics of the observed effect. Funato et al. observe a rapid but transient activation of TCF signaling by peroxide, peaking at 20 min, whereas we and others observe inhibition after prolonged periods of peroxide stress. However, a more intriguing possible explanation is suggested by the fact that Funato et al. employed NIH3T3 cells overexpressing Dishevelled. Shin et al. (36) showed that similar to our results ␤-catenin/TCF signaling is reduced after peroxide stress but that overexpression of Dishevelled could abrogate the effect of hydrogen peroxide. Thus, Dishevelled expression appears an important determinant in the sensitivity of TCF signaling toward oxidative stress. How this occurs precisely remains to be determined but could simply be a matter of signal strength. In that case, following Disheveled overexpression ␤-catenin levels would increase to such levels that binding of ␤-catenin to FOXO and TCF can be saturated simultaneously and thus competition cannot, or can no longer, occur.
Irrespective, these results further corroborate the notion that ␤-catenin/TCF signaling is tightly controlled by cellular oxidative stress. Similar to FOXO, both c-jun and HIF-1␣ are regulated by oxidative stress pathways. It can be concluded that under oxidative stress conditions it is apparently important for cells to diverge from TCF to FOXO and probably other (c-jun, HIF-1␣) stress-regulated transcriptional programs. Being the commonality, this suggests an important role for ␤-catenin in stress regulation; consistent with this idea, we observed previ-5 D. Hoogeboom, E. Voets, and B. Burgering, data not shown. ously in C. elegans that BAR-1, the C. elegans ␤-catenin homologue, regulates resistance against oxidative stress independent of its binding to POP-1, the TCF homologue in C. elegans (1).
Finally, our data establish a potential point of interaction between insulin/IGF1 signaling and Wnt signaling. In contrast to cellular oxidative stress, insulin does not substantially affect ␤-catenin FOXO interaction, at least as measured by co-immunoprecipitation. 6 Indeed, it has been shown that insulin/insulin-like growth factor can inhibit Wnt signaling at the level of ␤-catenin (37,38). Several other studies have explored the possibility that insulin/ insulin-like growth factor 1 would influence ␤-catenin by activation of protein kinase B/Akt and inhibition of GSK-3 (39,40). However, although these studies did indicate that insulin/insulin-like growth factor 1 to some extent may regulate TCF activity, the involvement of protein kinase B-regulated GSK-3 activity appeared unlikely. This has led to the suggestion that different pools of GSK-3 may exist and that these pools are restricted to either Wnt or insulin signaling (40,41). To fully appreciate the interaction between insulin and Wnt signaling clearly requires further studies.