Abstract
Wnt signaling stabilizes beta-catenin, which in turn influences the transcription of Wnt-responsive genes in conjunction with T-cell factor (TCF) transcription factors. At present, there are two models for the actions of beta-catenin. The conventional nuclear model suggests that beta-catenin acts in the nucleus to form a heterodimeric transcriptional factor complex with TCF, with TCF providing DNA-specific binding and the C and N termini of beta-catenin stimulating transcription. The alternative cytoplasmic model postulates that beta-catenin exports TCF from the nucleus to relieve its repressive activity or activates it in the cytoplasm. We have generated modified forms of beta-catenin and used RNA interference against endogenous beta-catenin to distinguish between these models in cultured mammalian and Drosophila cells. We show that the VP16 transcriptional activation domain can replace the C terminus of beta-catenin without loss of function and that the function of beta-catenin is compromised by fusion to a transcriptional repressor domain from histone deacetylase, favoring the direct effects of beta-catenin in the nucleus. Furthermore, membrane-tethered beta-catenin requires interaction with the adenomatous polyposis coli protein but not with TCF for its function, whereas untethered beta-catenin requires binding to TCF for its signaling activity. Importantly, by using RNA interference, we show that the signaling activity of membrane-tethered beta-catenin, but not free beta-catenin, requires the presence of endogenous beta-catenin, which is able to accumulate in the nucleus when stabilized by the binding of the beta-catenin degradation machinery to the membrane-tethered form. All of these data support a nuclear model for the normal function of beta-catenin.