Suppressor of Fused Negatively Regulates (cid:1) -Catenin Signaling*

Suppressor of fused (Su(fu)) is a negative regulator of the Hedgehog signaling pathway that controls the nu-clear-cytoplasmic distribution of Gli/Ci transcription factors through direct protein-protein interactions. We show here that Su(fu) is present in a complex with the oncogenic transcriptional activator (cid:1) -catenin and functions as a negative regulator of T-cell factor (Tcf)-de-pendent transcription. Overexpression of Su(fu) in SW480 (APC mut ) colon cancer cells in which (cid:1) -catenin protein is stabilized leads to a reduction in nuclear (cid:1) -catenin levels and in Tcf-dependent transcription. This effect of Su(fu) overexpression can be blocked by treatment of these cells with leptomycin B, a specific inhibitor of CRM1-mediated nuclear export. Overexpression of Su(fu) suppresses growth of SW480 (APC mut ) tumor cells in nude mice. These observations indicate that Su(fu) negatively regulates (cid:1) -catenin signaling and that CRM-1-mediated nuclear export plays a role in this regulation. Our results also suggest that Su(fu) acts as a tumor suppressor. fraction were compared between cultures by normalizing the (cid:1) -catenin level to tubulin or nuclear histone protein level. Xenograft Studies— SW480 cells, SW480 cells expressing Su(fu) at a low level (clone 2), and SW480 cells expressing Su(fu) at a high level (clone 7) were used to produce xenograft tumors in nude mice (The Jackson Laboratory, Bar Harbor, ME). One million cells from each cell type were injected into a separate subcutaneous location on the dorsum of the proximal hind limbs. Ten injection sites (five mice, two injection sites for each cell type in each mouse) for each cell type were analyzed. The size of the tumor was assessed by a blinded observer on a weekly basis.

The oncogenic transcriptional activator ␤-catenin is a major mediator in Wnt signaling (1)(2)(3)(4). A large multiprotein complex that includes APC 1 and axin normally facilitates the phosphorylation of ␤-catenin by GSK3␤. Phosphorylated ␤-catenin binds to the F-box protein ␤TrCP and is then modified by ubiquitination and subjected to proteasome-mediated protein degradation. When cells are exposed to the Wnt signal, ␤-catenin phosphorylation and its subsequent ubiquitination are blocked. ␤-Catenin is thus diverted from the proteasome; instead, ␤-catenin accumulates and translocates to the nucleus, where it interacts with members of the Tcf/Lef family of transcription factors and activates transcription of Wnt-responsive genes. In tumors, ␤-catenin degradation is blocked by mutations of APC, axin, or ␤-catenin itself. As a result, stabilized ␤-catenin enters the nucleus and ␤-catenin⅐Tcf complexes activate oncogenic target genes.
Nuclear translocation of ␤-catenin is of key importance in its ability to regulate transcription, yet little is known about the factors important in controlling the nuclear versus cytoplasmic distribution of ␤-catenin. ␤-Catenin lacks a nuclear import signal, and it docks to the nuclear membrane by a mechanism that is Ran-independent and does not require importins (5). Nuclear import of ␤-catenin is also independent of its association with the Tcf transcription factors because mutant forms of ␤-catenin that do not bind Tcf proteins can enter the nucleus (6). Microinjection studies show that ␤-catenin rapidly exits the nucleus, suggesting a role for nuclear export in the regulation of the intracellular distribution of ␤-catenin (7).
Several studies demonstrate that APC is a nucleo-cytoplasmic protein with export from the nucleus inhibited by LMB, a specific inhibitor of CRM1-mediated nuclear export (8 -10). CRM1, also called exportin-1, is an export karypopherin that binds to a leucine-rich nuclear export signal on its target protein and mediates nuclear-cytoplasmic trafficking of proteins as well as RNA through the nuclear pore. LMB binds directly to CRM1 and inactivates its nuclear export activity (11). In addition to the regulatory role of APC in ␤-catenin degradation, these studies suggest that APC promotes nuclear export of ␤-catenin (8 -10). However, several recent studies demonstrate that ␤-catenin is exported from the nucleus in a CRM1-and Ran-independent manner (12,13). Because the subcellular distribution of ␤-catenin is affected by many of its interacting proteins, it is possible that some of these ␤-catenin binding partners are regulated by CRM1-mediated nuclear export and that LMB treatment affects the nuclear localization of ␤-catenin indirectly.
Genetic screens in Drosophila first identified Su(fu) as a suppressor of the Fused kinase, a positive regulator of Hh signaling (14). Recent studies demonstrate that Su(fu) acts as a negative regulator of Hh signaling by directly interacting with the Ci/Gli zinc finger transcription factors, which are transducers of Hh signaling (15)(16)(17). Although the mechanisms by which Su(fu) functions remain unclear, Su(fu) is thought to control the nuclear-cytoplasmic distribution of Gli/Ci transcription factors through direct protein-protein interactions (18,19). In mammalian cells, overexpression of Su(fu) causes Gli1 to be concentrated in the cytoplasm. Inhibition of CRM1-dependent nuclear export by LMB treatment counteracts these effects of Su(fu) overexpression restoring the nuclear distribution of Gli1, thus suggesting that Su(fu) promotes CRM1-dependent nuclear export of Gli1 (20).
Here we report that Su(fu) and ␤-catenin are found in the same complex. In the human colon cancer SW480 cell line, overexpression of Su(fu) results in a reduction of nuclear ␤-catenin and Tcf-dependent transcription, suggesting that Su(fu) can act as a negative regulator of Wnt signaling. Studies with LMB treatment indicate that the ability of Su(fu) to regulate ␤-catenin is mediated by a CRM1-dependent nuclear export mechanism. We propose that Su(fu) can negatively regulate Wnt signaling by promoting nuclear export of ␤-catenin. Furthermore, we show that Su(fu) overexpression inhibits the growth of SW480 (APC mut ) colon cancer cells in nude mice, suggesting a role for Su(fu) as a tumor suppresser.

MATERIALS AND METHODS
Plasmids-A plasmid encoding mouse Su(fu) was constructed by subcloning the open reading frame of mouse Su(fu) cDNA (19) into the pCMV5␤ vector with an amino-terminal Myc tag or with an aminoterminal HA tag. A Myc-tagged Su(fu) construct used for stable trans-formation studies was also generated in the pcDNA3 vector. Su(fu) deletion mutants were generated by polymerase chain reaction mutagenesis and subcloned into the pCMV5␤ vector with an amino-terminal Myc epitope tag. The stabilized ␤-catenin mutant (Mut-␤-catenin) was generated using polymerase chain reaction mutatgenesis, converting the four serine or threonine phosphorylation sites (codons 33, 37, 41, and 45) to alanine using the full-length ␤-catenin gene (obtained from S. Hirohashi (21)), and was subcloned into the pCMV5␤ vector. All polymerase chain reaction-generated constructs were verified by sequencing. The GluGlu-tagged ⌬N89-␤-catenin (obtained from P. Polakis) was subcloned into the pCMV5␤ vector, and the wild type APC gene in a CMV vector (obtained from B. Vogelstein and K. Kinzler (22)) was subcloned into a FLAG-tagged vector.
Immunoprecipitations-Human 293T cells (kidney epithelial cell line) were passaged to 50% confluence, and transfection was carried out with DNA using the Superfect transfection reagent (Qiagen) according to the manufacturer's instructions. Cells were harvested 36 h after transfection. Lysates were prepared in immunoprecipitation buffer (50 mM Tris, pH 7.0, 150 mM NaCl, 0.5% Triton X-100, 1 mM EDTA, and protease inhibitors). Immunoprecipitations were performed using anti-␤-catenin antibody, anti-GluGlu antibody, anti-Myc antibody, or anti-HA antibody, and protein G-agarose beads. The immunoprecipitates were separated on 8% SDS-polyacrylamide gel electrophoresis, and Western analysis was performed using the indicated primary and corresponding secondary horseradish peroxidiase-conjugated antibodies (Jackson ImmunoResearch). Enhanced chemiluminescence detection (Pierce) was performed according to the manufacturer's instructions. The ␤-catenin antibody was obtained from BD Transduction Laboratory; the Su(fu) antibody was a rabbit polyclonal antibody raised against a synthesized peptide fragment from amino acid residues 440 -457 of mSu(fu); the anti-GluGlu and anti-HA antibodies were obtained from Babco; and the anti-Myc antibody was obtained from Santa Cruz Biotechnology. Immunoprecipitation of native proteins was also performed using lysates from the SW480 colon cancer cell line in an identical manner.
Tcf Transcriptional Activation-Cells plated on 35-mm dishes were transiently transfected in triplicate with 1 g of the luciferase reporter construct pTOPFLASH or pFOPFLASH (23). In addition, some cell dishes were also co-transfected with 1 g of ⌬N89-␤-catenin, 1 g of mutant ␤-catenin, 1 g of the full-length APC gene, or empty vector controls. Where indicated, some of the cells were treated with 20 ng/ml LMB (obtained from M. Yoshida) for 90 min or 16 h before measuring Tcf transcriptional activation. Superfect transfection reagent (Qiagen) was utilized according to the manufacturer's instructions, and in all cases, a Rous sarcoma virus ␤-galactosidase expression vector was used as a control for transfection efficiency. Cells were harvested 24 h after transfection, and luciferase enzyme activity was measured using a luminometer, and normalized to ␤-galactosidase activity. A ratio of the normalized pTOPFLASH/pFOPFLASH luminescence was calculated for each cell dish. The means and standard deviations, determined for each cell type and transfection condition, were compared using the two-way t test.
Subcellular Localization-SW480 cells were transfected with or without appropriate expression constructs. Where indicated, 20 ng/ml LMB was added to the cells 90 min or 16 h before harvesting. After transfection with Myc-Su(fu) in a pcDNA3 expression vector, stable SW480 transformants were isolated after 8 days of G418 selection. The clones were then expanded and analyzed for Myc-Su(fu) expression using Western blot analysis. Two clones, clone 2 and clone 7, were used for this study; clone 7 expresses Myc-Su(fu) at a level about 5-fold higher than clone 2. The cells were fixed for 10 min in 4% formaldehyde in phosphate-buffered saline and then permeabilized in methanol for 2 min. Endogenous ␤-catenin was detected by monoclonal anti-␤-catenin antibody (BD Transduction Laboratory) and fluorescein isothiocyanateconjugated goat anti-mouse IgG (Jackson ImmunoResearch). Myc-Su(fu) and its mutants were detected by polyclonal anti-Myc antibody (Santa Cruz Biotechnology) and rhodamine-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch). For fluorescence microscopy, a Zeiss confocal microscope was used. Nuclei were isolated from cell cultures using a process of freezing, thawing, and lysing as reported by Wood and Earnshaw (24). Western analysis for ␤-catenin, actin, and histone was perfomed on the cytoplasmic and nuclear fractions as previously reported (25). The ␤-catenin protein levels in the cytoplasmic or nuclear fraction were compared between cultures by normalizing the ␤-catenin level to tubulin or nuclear histone protein level.
Xenograft Studies-SW480 cells, SW480 cells expressing Su(fu) at a low level (clone 2), and SW480 cells expressing Su(fu) at a high level (clone 7) were used to produce xenograft tumors in nude mice (The Jackson Laboratory, Bar Harbor, ME). One million cells from each cell type were injected into a separate subcutaneous location on the dorsum of the proximal hind limbs. Ten injection sites (five mice, two injection sites for each cell type in each mouse) for each cell type were analyzed. The size of the tumor was assessed by a blinded observer on a weekly basis.

␤-Catenin and Su(fu) Are Present in the Same Complex-To
determine whether Su(fu) interacts with ␤-catenin, epitopetagged Su(fu) was transiently expressed, either alone or in combination with a stabilized form of ␤-catenin (⌬N89-␤-catenin, which lacks the first 89 amino acid residues important for protein degradation (26)), in 293 cells. As shown in Fig. 1, Myc-tagged Su(fu) co-precipitates with endogenous ␤-catenin (lane 7) as well as GluGlu-tagged ⌬N89-␤-catenin (lane 11), when Myc-tagged Su(fu) is immunoprecipitated with an anti-Myc antibody. Conversely, Myc-Su(fu) is also present in the immunocomplex of ␤-catenin precipitated with either an anti-␤-catenin or anti-GluGlu antibody (data not shown). Interactions between endogenous Su(fu) and ␤-catenin were also verified using a polyclonal antibody raised against Su(fu) in co-immunoprecipitation experiments in 293 cells (Fig. 1, lane 5) and the human colon cancer SW480 cell (data not shown). These results show that Su(fu) interacts with ␤-catenin and that this interaction occurs in the absence of the first 89 amino acids of ␤-catenin, which have previously been shown to be required for binding of the F-box protein, ␤TrCP.
Su(fu) Overexpression Inhibits Tcf-dependent Transcriptional Activation-In response to the activation of the Wnt signaling pathway, stabilized ␤-catenin translocates to the nucleus where it regulates transcription of Wnt-responsive genes via interactions with the Tcf/Lef family of transcription factors. To determine whether Su(fu)-␤-catenin interactions have any effects on the transcriptional function of ␤-catenin, a Tcf-luciferase reporter assay was employed. Tcf-dependent transcriptional activation was measured using the pTOP-FLASH (luciferase reporter with optimized Tcf-binding sites) and pFOPFLASH (luciferase reporter with mutant Tcf-binding sites) constructs in transient expression assays (23). In NIH3T3 cells, where endogenous ␤-catenin levels are low, only weak Tcf-dependent transcriptional activation was detected (Fig. 2). Expression of either of the stabilized forms of ␤-catenin (either ⌬N89-␤-catenin or Mut-␤-catenin) in these cells resulted in a 2-3ϫ increase in Tcf-dependent transcription, whereas Su(fu) itself had no obvious effects. Co-expression of Su(fu) with either of the stabilized forms of ␤-catenin resulted in a slight but significant reduction of Tcf-dependent transcription. The inhibition of Tcf-dependent transcription by Su(fu) is weaker than that by APC but is comparable with the level of inhibition as reported in several ␤-catenin-interacting proteins, such as duplin, pontin52, and reptin52 (27,28). These observations suggests that Su(fu)-␤-catenin interactions result in a down-regulation of Tcf-dependent transcription.
Su(fu) Overexpression Reduces Nuclear ␤-Catenin Levels-To further investigate the action of Su(fu) on Tcf-dependent transcription and ␤-catenin levels, we studied the human colon cancer cell line SW480, which exhibits elevated levels of  (E and H). The amino-terminal Su(fu) deletion mutant has a subcellular location similar to the wild type Su(fu) (K). Although wild type Su(fu) decreases nuclear ␤-catenin, the carboxyl-terminal Su(fu) deletion mutants lose the ability to decrease endogenous nuclear ␤-catenin. Merged views show co-localization of Su(fu) and ␤-catenin. Panel M shows Tcf transcriptional activation in SW480 cells with overexpression of Su(fu) or its deletion mutants. An asterisk next to the pTOPFLASH/ pFOPFLASH ratio indicates a statistically significant difference compared with transfection with an empty vector (p Ͻ 0.05 t test). Su(fu) expression decreases Tcf transcriptional activation, but the carboxylterminal deletion mutants of Su(fu) do not alter Tcf transcriptional activation.
␤-catenin because of a truncating mutation in APC (29). The subcellular distribution of ␤-catenin in SW480 cells was examined by confocal microscopy. In SW480 cells, high levels of ␤-catenin were found in both the nucleus and the cytoplasm. Strikingly, SW480 cells overexpressing Su(fu) showed a drastic reduction of nuclear staining of ␤-catenin (Fig. 3, A-C). This result suggests that Su(fu) overexpression reduces nuclear ␤-catenin levels, which lead to a down-regulation of Tcf-dependent transcription.
The effect of Su(fu) on the regulation of nuclear ␤-catenin was examined by using several Su(fu) deletion mutants. In SW480 cells, wild type Su(fu) was predominantly cytoplasmic (Fig. 3B), whereas, in contrast, Su(fu) mutants with carboxylterminal deletions showed significant nuclear accumulation (Fig. 3, E and H). Interestingly, these Su(fu) mutants have lost the ability to reduce nuclear ␤-catenin level (Fig. 3, F and I) and to down-regulate Tcf-dependent transcription (Fig. 3 M). In contrast, a Su(fu) mutant lacking only the amino-terminal region has activities similar to the wild type protein (Fig. 3, L and  M). Although these mutant proteins exhibit distinct activities, they all retain their ability to form a complex with ␤-catenin as assayed by immunoprecipitation (data not shown). Taken together, these results indicate that the carboxyl-terminal region of Su(fu) is required for reduction of nuclear ␤-catenin levels and thus down-regulation of Tcf-dependent transcription.

LMB Treatment Counteracts the Effect of Su(fu) on Reducing the Nuclear ␤-Catenin Level and Inhibiting Tcf-dependent
Transcription-Because APC is mutated in SW480 cells, we were intrigued by the possibility that Su(fu) might regulate the nuclear-cytoplasmic distribution of ␤-catenin level in an APCindependent manner. To address this, we treated transiently transfected SW480 cells with LMB, a specific inhibitor of CRM1-mediated nuclear export (11,30) (Fig. 4). Strikingly, LMB treatment restored a high level of ␤-catenin staining in Merged views show co-localization of Su(fu) and ␤-catenin. Panel Q shows the effects of LMB treatment on Tcf transcriptional activation. LMB treatment results in a significant increase (p 0.05 t test) in the pTOPFLASH/pFOPFLASH ratio in Su(fu) overexpressing SW480 cells (indicated by an asterisk). Panel R shows ␤-catenin protein level in nuclear and cytoplasmic cell fractions as determined by Western analysis. Western analysis for tubulin or nuclear histone from teh same blot is shown as a loading control. This shows a decrease in nuclear ␤-catenin with expression of Su(fu), and an increase with LMB treatment. the nucleus of Su(fu)-overexpressing cells, suggesting that the reduction of nuclear ␤-catenin level is mediated through a CRM1-mediated nuclear export mechanism (Fig. 4, B-D and  F-H).
To further examine the CRM1-dependence of Su(fu) function, stable SW480 transformants overexpressing Su(fu) were established. As in our transient expression studies, Su(fu) overexpression resulted in a reduction of nuclear ␤-catenin levels (Fig. 4, J-L, and data not shown) and down-regulation of Tcf-dependent transcription in the stable transformants (Fig.  4Q). LMB treatment counteracted the effects of Su(fu) overexpression in these SW480 transformants (Fig. 4, N-Q, and data  not shown). Similar results were observed in cells with LMB treatment for 90 min or 16 h. The subcellular distribution of Su(fu) itself is regulated by nuclear export, as Su(fu) could readily be detected in the nucleus of LMB-treated cells, suggesting that it is transiently present in the nucleus and actively exported (Fig. 4, G and O). These results suggest that the subcellular location of Su(fu) regulates ␤-catenin-mediated Tcf-dependent transcription.
To determine the subcellular distribution of ␤-catenin in a more quantitative manner, we examined protein levels in cytoplasmic or nuclear fractions using Western analysis. Extracts from SW480 cells and SW480 cells expressing Su(fu) (clone 7) with or without LMB treatment were separated into cytoplasmic and nuclear fractions. The purity of the fractions was determined by Western analysis using actin and histone as markers for cytoplasmic and nuclear fractions, respectively. Consistent with the results of confocal microscopy studies, SW480 cells with Su(fu) overexpression had a lower level of nuclear ␤-catenin, and LMB treatment increased the level of nuclear ␤-catenin in these cells to those of the parental SW480 cells (Fig. 4R). Together, these results indicate that Su(fu) reduces nuclear ␤-catenin levels through a CRM1-dependent nuclear export mechanism.
Differential Effects of LMB on the Consequences of Su(fu) versus APC Overexpression in SW480 Cells-Because Su(fu) can promote nuclear export of ␤-catenin in the absence of normal APC function, we compared the effects of APC overexpression versus Su(fu) overexpression in SW480 cells (Fig. 5). In SW480 cells, APC overexpression resulted in a drastic reduction of Tcf-dependent transcription, whereas the inhibitory effect of Su(fu) overexpression was significantly less pronounced (Fig. 5A). Consistent with the notion that Tcf-dependent transcription is regulated by the level of nuclear ␤-catenin, a low level of ␤-catenin was found in Su(fu)-expressing cells (Fig. 5, B-D), whereas very little or no ␤-catenin staining could be detected in APC-expressing cells (Fig. 5, H-J).
Since we demonstrated that reduction of nuclear ␤-catenin level by Su(fu) overexpression was abolished by treating the cells with LMB (Fig. 5, E-G), we examined the consequences of simultaneous overexpression of APC and Su(fu) in SW480 cells. The effects were additive, with APC and Su(fu) expression resulting in an almost complete elimination of ␤-catenin (Fig.  5, O-Q). After LMB treatment, ␤-catenin remained undetectable in the nucleus of the APC-and Su(fu)-overexpressing cells (Fig. 5, S-U), suggesting that the role of APC in promoting ␤-catenin protein degradation is dominant over that of Su(fu). In this case, ␤-catenin is mostly degraded in the cytoplasm through the action of wild type APC, and very little ␤-catenin can be translocated to the nucleus. These results are consistent with the notion that Su(fu) acts to regulate the activity and localization of ␤-catenin after nuclear translocation of ␤-catenin.
Su(fu) Overexpression Attenuates Tumor Formation by SW480 Cells in Nude Mice-To determine whether Su(fu) al-ters the oncogenic function of ␤-catenin, the size of xenograft tumors formed after subcutaneous injection of SW480 cells into nude mice was compared using SW480 cells with or without overexpression of Su(fu). As shown in Fig. 6, Su(fu) overexpression resulted in a significant reduction in the rate of growth of these tumors. There was also a correlation between the level of Su(fu) expression and the effect on tumor growth; Clone 7, which expresses higher levels of Su(fu), showed less tumor growth than clone 2, which has lower Su(fu) expression. These results indicate that Su(fu) overexpression can suppress the oncogenic functions of ␤-catenin. DISCUSSION We have shown that Su(fu) and ␤-catenin reside in the same complex and that Su(fu) negatively regulates Tcf-dependent transcription by reducing nuclear ␤-catenin levels. These inhibitory activities of Su(fu) are sensitive to LMB treatment indicating that the reduction of nuclear ␤-catenin is accomplished through a CRM1-mediated nuclear export mechanism. Interestingly, although APC is implicated in both nuclear export and cytoplasmic degradation of ␤-catenin, Su(fu) promotes ␤-catenin nuclear export in SW480 cells, which lack normal APC function. Thus, our results indicate that Su(fu) regulates Wnt signaling through APC-independent mechanism in which CRM1-mediated nuclear export plays a role.
We also showed that the action of APC is dominant over that of Su(fu); LMB treatment failed to block the reduction of nuclear ␤-catenin level in cells overexpressing both APC and Su(fu) (see Fig. 5). In normal cells, once ␤-catenin is synthesized in the cytoplasm, it will be targeted for degradation by the destruction complex; this targeting involves APC, axin, and GSK3␤ (1)(2)(3)(4). Although a role for Su(fu) in the cytoplasmic compartment cannot be excluded, our results suggest that the major role of Su(fu) is after nuclear translocation of ␤-catenin. In APC-overexpressing SW480 cells, because most ␤-catenin will be degraded in the cytoplasm, and very little ␤-catenin is imported into the nucleus; the role of Su(fu) in ␤-catenin signaling thus becomes minimal.
The accumulation of Su(fu) in the nucleus of LMB-treated cells suggests that the intracellular distribution of Su(fu) itself is regulated by CRM1-mediated nuclear export. We propose that there is a nuclear-cytoplasmic exchange of Su(fu) itself and that Su(fu) interacts with ␤-catenin in the nucleus, facilitates the nuclear export of ␤-catenin, and eventually leads to a down-regulation of ␤-catenin-mediated Tcf-dependent transcription. Because mutant forms of Su(fu) lacking the carboxylterminal region are enriched in the nucleus, this carboxylterminal region may be essential for the efficient nuclear export of Su(fu). Elevated ␤-catenin-mediated transcription in this case indicates that either these same regions of Su(fu) are important for the ability of Su(fu) to regulate ␤-catenin mediated transcription or that nuclear export of Su(fu) is necessary for it to reduce nuclear ␤-catenin levels. In this latter possibility, Su(fu) might act in a complex to export ␤-catenin from the nucleus.
What is the role of Su(fu) in ␤-catenin signaling in vivo? Numerous studies have suggested that nuclear localization of ␤-catenin is a key step in Wnt signaling and that the intracellular localization of ␤-catenin is controlled by the availability and affinity of its binding partners (1)(2)(3)(4). We propose that Su(fu) is one of these binding partners and that Su(fu) functions to control the level of nuclear ␤-catenin in cells receiving Wnt signals. As ␤-catenin signaling is over-activated in many benign and malignant tumors, Su(fu) would thus be expected to act as a tumor suppressor. Consistent with this notion, we find that overexpression of Su(fu) in SW480 cells suppresses tumor growth in nude mice. It will now be important to examine whether inactivating mutations of Su(fu) are found in tumors with activated ␤-catenin signaling.
The primary focus here has been on the role of Su(fu) in ␤-catenin signaling. However, it is also clear that Su(fu) acts as a negative regulator of Hh signaling in both flies and mammals. Several studies have suggested that Su(fu) might possess multiple functions controlling the activities of Ci/Gli transcription factors, including cytoplasmic sequestration (17-20, 31, 32) and increase of DNA binding (16). The present study has revealed a novel role for Su(fu) in the regulation of ␤-catenin signaling through CRM1-mediated nuclear export. Because Ci/Gli transcription factors possess nuclear export signals and LMB treatment has been shown to increase nuclear localization of Ci in Drosophila (33) and to inhibit cytoplasmic sequestration of Gli1 (20), we propose that Su(fu) also regulates the activity of ␤-catenin and Ci/Gli transcription factors through a similar CRM1-mediated mechanism. In both cases, it is possible that Su(fu) acts to regulate the nuclear location of these transcriptional regulators through an export-related mechanism.
Finally, the observation that Su(fu) is involved in the regulation of ␤-catenin signaling provides further evidence that multiple levels of cross-talk exist between Wnt and Hh signaling during development and cancer. In Drosophila, the F-box protein and ␤-TrCP homolog, Slimb, is known to regulate the degradation of Armadillo (␤-catenin homolog) and Ci (34). The transcriptional cofactor p300 can interact with both ␤-catenin and Ci/Gli, promoting their transcriptional activities (35)(36)(37)(38)(39). Our results suggest that Su(fu) may regulate the activities of both ␤-catenin and Gli1 through a CRM1-mediated nuclear export mechanism.
FIG. 6. Su(fu) overexpression suppresses growth of xenograft tumors in nude mice. SW480 cells and two cell clones with Su(fu) overexpression were injected subcutaneously into nude mice, and the size of their xenograft tumors was measured 1 and 2 weeks after injection. The mean and standard deviation of the tumor size are shown. Both clone 2 (low Su(fu) overexpression) and clone 7 (high Su(fu) overexpression) showed significantly lower rates of tumor growth (p Ͻ 0.05 t test).