Reduction of β-Catenin/T-Cell Transcription Factor Signaling by Aspirin and Indomethacin Is Caused by an Increased Stabilization of Phosphorylated β-Catenin

Introduction

Alterations in pathways regulating β-catenin degradation are very common in the pathogenesis of many human cancers (reviewed in Ref. 1). Particularly in colorectal carcinogenesis, enhanced stability of β-catenin resulting in its nuclear accumulation is regarded to be crucial for initiation of this process. Nuclear β-catenin serves as a coactivator of TCF²/lymphoid enhancer-binding factor-induced transcription of target genes that play important roles in tumor progression (2). The localization and thus the function of β-catenin are thought to be regulated by tyrosine and S/T phosphorylation (2), thereby separating three different subcellular β-catenin pools: membrane bound; cytosolic; and nuclear. Four serine/threonine residues (S33, S37, T41, and S45) are conserved from Drosophila to human and conform to the consensus phosphorylation site for GSK-3β (3). Recently, it has been suggested that β-catenin is targeted for ubiquitin-dependent degradation by a dual kinase mechanism. CKIα-mediated phosphorylation of S45 is required for subsequent phosphorylation at β-catenin residues T41, S37, and S33 by GSK-3β (4). In unstimulated resting cells, GSK-3β is a constitutively active enzyme that is negatively regulated, i.e., by the Wnt pathway (5) and by Akt-induced phosphorylation (6), both resulting in accumulation of β-catenin. Phosphorylation and degradation of β-catenin is triggered by a multiprotein complex consisting of GSK-3β, adenomatous polyposis coli, axin (7), and β-TrCP, an F-box component of the E3 ubiquitin ligase complex (8). The significance of β-catenin phosphorylation is underscored by the observation that mutations in the GSK-3β consensus phosphorylation site frequently occur in human colorectal cancer and several other malignancies (reviewed in Ref. 1). Accordingly, enormous efforts are made to detect specific inhibitors of the protein components of the β-catenin-pathway(s) for development of target-directed therapies.

Epidemiological and pharmacological evidence suggest the potential use of NSAIDs in cancer chemoprevention (9, 10). The cancer protective activity of NSAIDs generally has been attributed to direct inhibition of the enzyme cyclooxygenase 2, which is commonly overexpressed in tumors. However, additional mechanisms likely contribute to the antineoplastic effects of NSAIDs (11–14). Recently, we have shown that the proapoptotic NSAIDs aspirin and indomethacin down-regulate β-catenin/TCF signaling in colorectal cancer cell lines (15) at pharmacologically relevant doses that are comparable with levels measured in plasma from patients given a short course of analgesic doses (16, 17). Within 24 h, the transcriptional activity of the β-catenin/TCF complex was strongly inhibited by these NSAIDs without affecting the nuclear translocation or the total amount of β-catenin. This led us to investigate whether the reduced signaling activity of β-catenin in response to NSAIDs was a result of its enhanced S/T-phosphorylation.

Using a panel of antibodies that specifically recognize β-catenin phosphorylation sites, we demonstrate that both aspirin and indomethacin indeed induce S/T-phosphorylation starting 1–6 h after drug administration in colorectal cancer cells. The phosphorylation preceded down-regulation of β-catenin-signaling activity, thus suggesting a causal relationship.

Materials and Methods

Cell Culture and Reagents.

Human colorectal cancer cell lines (SW480, SW948, HCT116, LS174T, HT-29, Colo320, CX-2, and LoVo) and the human embryonic kidney cell line 293 were derived from the tumor collection of the German Cancer Research Center (Tumorbank; DKFZ, Heidelberg, Germany). Cells were grown in RPMI 1640 (Life Technologies, Inc.) supplemented with 10% FCS, 100 units/ml penicillin, and 100 μg/ml streptomycin at 37°C in a humidified atmosphere of 5% CO₂.

LiCl, calyculin A, aspirin, and indomethacin were supplied by Sigma (Taufkirchen, Germany). Stock solutions of 1 m aspirin or 100 mm indomethacin were prepared in acetone and added to the cell culture media at the indicated concentrations. The pH was adjusted to 7.0–7.5 by TRIS base. A 1 m stock solution of LiCl was prepared in RPMI 1640, and a 50 μm calyculin A stock solution was dissolved in DMSO.

Transfection and Reporter Gene Assays.

Transient transfections using the TCF-4/β-catenin-responsive Topflash/Fopflash reporter constructs (Hans Clevers; University of Utrecht, Utrecht, the Netherlands) were performed as described previously (15). NSAIDs were added 16 h after transfection, and the cells were harvested after 30 min, 60 min, 6 h, and 24 h of NSAID treatment as indicated in the figures. Lysates were prepared in Tris-PO₄-luciferase buffer and collected for assays of luciferase and galactosidase activity, respectively, which were performed as described previously (18).

Immunoblotting.

For analysis of growth factor-mediated effects, cells were serum-starved for 24 h through incubation in serum-free RPMI 1640 before adding 10% FCS, aspirin, indomethacin, calyculin A, or LiCl as indicated in the figures. Equal cell numbers (10⁶) of serum-deprived and nondeprived cells were collected in 100 μl of sample buffer [0.25 m Tris-Cl (pH 6.8), 20% glycerin, 4% SDS, and 10% β-mercaptoethanol] at different time intervals and homogenized by sonification before separation by SDS-PAGE and electroblotting. For detection of proteins and phosphorylation sites, antibodies against β-actin (ICN Biomedicals, Aurora, OH), β-catenin (Transduction Laboratories), P33.37.41-catenin, P41.45-catenin, P9-GSK-3β, IκB, and P32.36-IκB (Cell Signaling Technology, Beverly, MA) were used according to the manufacturer’s instructions. Secondary antibodies, goat antirabbit or rabbit-antimouse IgG horseradish peroxidase conjugates, were purchased from Jackson ImmunoResearch, Dianova (Hamburg, Germany). The immunoblots were visualized by enhanced chemiluminescence with SuperSignal West Femto-Substrate (Pierce, Rockford, IL; for detection of weak phosphorylation signals) or enhanced chemiluminescence (Amersham Biosciences Europe, Freiburg, Germany).

Results

S/T-Phosphorylation of β-catenin Is Induced by Aspirin and Indomethacin and Precedes Down-Regulation of β-Catenin/TCF Transcriptional Activity.

On the basis of our previous observations that β-catenin/TCF signaling is down-regulated by aspirin and indomethacin within 24 h in a dose-dependent manner in colorectal cancer cell lines, we asked whether the reduced signaling activity was a result of an increase in β-catenin phosphorylation. β-Catenin/TCF signaling is constitutively activated in SW480 colorectal cancer cells because of high nuclear β-catenin expression levels (19). We compared the aspirin response of SW480 cells with that of the embryonic kidney cell line 293, expressing low levels of β-catenin and exhibiting low endogenous β-catenin/TCF activity (20) by Western blot analysis using phosphorylation site-specific antibodies. According to our (15) and others’ (17) earlier results, we performed the experiments with 5 mm aspirin, a concentration that is commonly used for in vitro experiments in colorectal cancer cells and was reported to be close to those achieved in plasma of patients upon long-time use (16). Fig. 1 A depicts the time-dependent effect of aspirin on S/T-phosphorylation of β-catenin in SW480 and 293 cells. Indeed, a clear induction of phosphorylated β-catenin was found in SW480 cells within 24 h in the presence of aspirin. In untreated SW480 cells, a low endogenous level of phospho-S33/S37/T41-catenin was detected. Addition of aspirin was followed by an initial 30 min period of reduced phospho-S33/S37/T41-catenin. Starting 1 h after treatment, the phospho-S33/S37/T41-catenin level increased at 6 h and remained unchanged until 24 h of exposure to aspirin (Fig. 1A, top panel). Phospho-T41/S45-catenin was absent in untreated SW480 cells and became first detectable at 6 h, remaining unchanged until 24 h of aspirin treatment. In contrast, aspirin did not display any appreciable impact on β-catenin phosphorylation in 293 cells, however, a weak signal corresponding to phospho-T41/S45-catenin could be detected after 24 h of treatment (Fig. 1 A, bottom panel).

In SW480 and 293 cells, calyculin A, a potent S/T-phosphatase inhibitor, which was reported to induce hyperphosphorylation of β-catenin (21), clearly stabilized S33, S37, T41, and S45 phosphorylation, implying that there are constitutively active kinases stimulating β-catenin phosphorylation on these residues in both cell types.

The total level of β-catenin did not decrease upon treatment with aspirin. However, we had demonstrated in a recent report that aspirin and indomethacin down-regulate β-catenin/TCF signaling in a dose-dependent manner in colorectal cancer cell lines (15). To determine whether S/T-phosphorylation of β-catenin stands in causal relationship with the reduced transcription of target genes, parallel samples of the SW480 and 293 cells used for Western blot analysis were transiently transfected with a β-catenin/TCF-responsive luciferase reporter construct (Fig. 1B). In SW480 cells, activation of the reporter gene was reduced by 25% within 1–6 h of exposure to aspirin compared with untreated cells, which were harvested at the same time points. After 24 h, aspirin treatment resulted in a 94% inhibition of luciferase activation. Because the aspirin-induced β-catenin phosphorylation preceded down-regulation of β-catenin-signaling activity by several hours, we assume that this phosphorylation is more likely a cause than a consequence. In 293 cells, the endogenous β-catenin/TCF-signaling activity was only 1/15 of that in SW480 cells (1.7 in 293 versus 26.8 in SW480 cells). As expected from the low β-catenin level and its undetectable phosphorylation status in 293 cells, aspirin treatment displayed a much weaker inhibition on the reporter activity, which peaked at 24 h with a 28% reduction. Taken together, the impact of aspirin on the phosphorylation status of β-catenin correlated well with the following signaling activity.

The aspirin-induced increase of β-catenin phosphorylated at the T41 and S45 residues was also noted in aspirin-treated SW948 colorectal cancer cells, although it was not detectable at S33, S37 in these cells (Fig. 2A and data not shown). Accordingly, β-catenin/TCF-signaling activity was reduced to an intermediate level of 40% within 24 h of aspirin treatment (data not shown). We additionally examined whether indomethacin, which like aspirin, was shown to attenuate β-catenin-signaling activity (15), could also induce β-catenin phosphorylation. Indeed, indomethacin treatment caused a time-dependent increase in the phospho-β-catenin level, as exemplarily determined by Western blot analysis in SW948 cells with both phosphospecific antibodies, antiphospho-S33/S37/T41-catenin, and antiphospho-T41/S45-catenin (Fig. 2B). In SW948 cells, indomethacin displayed a substantially stronger effect on increasing the phospho-catenin level than aspirin and peaked earlier at 1 h after treatment. Consistent with the assumption that GSK-3β participates in the observed phosphorylation, LiCl, a potent GSK-3β inhibitor could reduce both aspirin- and indomethacin-mediated increase of the phospho-β-catenin level. When SW948 cells were pretreated with 20 mm LiCl for 30 min and washed before addition of NSAIDs, the phosphorylation level increased later than in cells that were not pretreated with LiCl (Fig. 2). We conclude from these data that GSK-3β is indeed involved in the NSAID-induced β-catenin phosphorylation.

The Phosphorylation Consensus Site Shared by β-Catenin and IκB Is Differently Modulated by Aspirin.

To further explore the mechanism responsible for the NSAID-induced increase of phospho-β-catenin, we next compared the β-catenin phosphorylation pattern with that of the nuclear factor-κB-regulating protein IκB, the degradation of which was already described to be modulated by aspirin (22). Two of the serines, which we could show to be phosphorylated in β-catenin in response to NSAIDs, are located within a motif that is almost identical to a six-amino acid sequence in IκB (Ref. 23; Fig. 3A). Like β-catenin, IκB is targeted for ubiquitination and proteasomal degradation upon serine phosphorylation (24). We therefore asked whether IκB phosphorylation at S32/S36 is modulated by aspirin in a similar way like β-catenin. To account for growth factors that might also influence the phosphorylation status of signaling molecules, we compared the response to aspirin of serum-starved cells versus cells grown in the presence of FCS. When SW480 cells were grown in serum-free medium, β-catenin phosphorylation at residues 33, 37, and 41 was detectable in untreated and aspirin-treated samples at any time point. Nevertheless, the phosphorylation level of all four residues was increased by aspirin in the absence as well as in the presence of serum at 6 and 24 h after treatment, without affecting the total amount of β-catenin (Fig. 3B, Lanes 5–8). In contrast, aspirin had no detectable effect on IκB phosphorylation in SW480 cells, neither in the presence nor absence of serum. Furthermore, the total amount of IκB was not affected by aspirin in these cells. Phospho-S32-IκB was only detectable upon stabilization with calyculin A, which resulted in subsequent IκB degradation (Fig. 3B, Lanes 3, 4, and 15).

On the contrary, aspirin transiently induced IκB phosphorylation within 30 min but rapidly declined in serum-starved 293 cells (Fig. 3C, Lane 5). IκB phosphorylation in 293 cells was also induced by FCS (Fig. 3C, Lane 9) and transiently increased upon aspirin treatment within 1 h (Fig. 3C, Lanes 10 and 11). The aspirin-induced phosphorylation did, however, not result in a similar extent of IκB degradation such as the stabilization of IκB phosphorylation by the phosphatase inhibitor calyculin A (Fig. 3C, Lanes 3, 4, and 15) or by induction through LiCl (Fig. 3C, Lane 14). Unlike in 293 cells grown in the presence of FCS and in contrast to the effect in SW480 cells, the total amount of β-catenin was clearly reduced by aspirin after 24 h of treatment when 293 cells were grown in the absence of serum (Fig. 3C, Lane 8). In summary, these experiments show that, although IκB and β-catenin share a common phosphorylation consensus site at the NH₂ terminus, aspirin exerts different effects on phosphorylation of these sites and the subsequent degradation of the corresponding molecules.

Participation of GSK-3β in Aspirin-induced Increase of Phospho-β-Catenin Levels.

The finding that the NSAID-induced phosphorylation of β-catenin could be inhibited by LiCl (Fig. 2) suggested involvement of GSK-3β in this process. In contrast to other S/T kinases, GSK-3β is constitutively active in unstimulated resting cells and is negatively regulated through association with other proteins or through phosphorylation by upstream kinases in several signaling pathways (25). Recently, it was reported that phosphorylation of GSK-3β by AKT results in its inactivation and subsequent accumulation of unphosphorylated β-catenin in macrophages (5, 26). We therefore asked whether the NSAID-induced β-catenin phosphorylation in colon cancer cells is mediated via altered GSK-3β activity, first focusing on the phosphorylation state of GSK-3β. In both serum-starved and in FCS-stimulated SW480 cells, GSK-3β was basically not phosphorylated (Fig. 3B, Lanes 1 and 9). The GSK-3β inhibitor LiCl weakly induced GSK-3β phosphorylation, indicating its inactivation (Fig. 3B, Lanes 2 and 14), and the phosphorylation was strongly stabilized by the phosphatase inhibitor calyculin A (Fig. 3B, Lanes 3, 4, and 15). Unexpectedly, phospho(S9)-GSK-3β was also clearly increased by aspirin in SW480 cells at 6 and 24 h after treatment, independent from growth factor presence (Fig. 3B, Lanes 7, 8, 12, and 13). Therefore, the GSK-3β phosphorylation/inactivation run in parallel with that of β-catenin, arguing against the presumption that GSK-3β activation might be the direct cause of aspirin-induced β-catenin phosphorylation in SW480 cells.

In contrast to SW480 cells, a clear basal level of phospho-S9-GSK-3β was detected in untreated serum-starved and in FCS-stimulated 293 cells (Fig. 3C, Lanes 1 and 9). As expected, LiCl additionally increased (Fig. 3C, Lanes 2 and 14) and calyculin A strongly stabilized S9-GSK-3β phosphorylation (Fig. 3C, Lanes 3, 4, and 15). Unlike in SW480 cells, the level of S9-GSK-3β phosphorylation time dependently decreased in serum-starved 293 cells upon aspirin treatment, indicating GSK-3β activation (Fig. 3C, Lanes 1 and 5–8). When aspirin was added to 293 cells in the presence of serum, GSK-3β was initially inhibited by a strong phosphorylation within 30 min, which subsequently decreased within 24 h to a level below that of untreated cells (Fig. 3C, Lanes 9–13). In both, serum-starved and FCS-treated 293 cells, this GSK-3β activation, however, did not result in detectable levels of phospho-catenin. Thus, the phosphorylation state and activity of GSK-3β in response to aspirin appears to be cell type specific. Immunoblots of six additional colon cancer cell lines confirmed that aspirin induces phosphorylation of GSK-3β in this cell type without affecting the total amount of β-catenin (Fig. 4A). Therefore, although inhibition of β-catenin phosphorylation by LiCl indicates involvement of GSK-3β (Fig. 2), the aspirin-induced increase of phospho-β-catenin in colorectal cancer cells seems to be mediated by a GSK-3β-independent mechanism, at least in part.

Discussion

In this study, we addressed the mechanisms underlying the aspirin- and indomethacin-caused reduction of β-catenin/TCF signaling in colorectal cancer cells. By using phosphospecific antibodies, we followed the phosphorylated intermediate in degradation of β-catenin in cultured cells upon treatment with aspirin or indomethacin. Examination of SW480 and SW948 colorectal cancer cells indicated that NSAID treatment increased the amount of phosphorylated β-catenin in colon cancer cells in a time-dependent manner, whereas β-catenin remained unaffected in human embryonic kidney 293 cells. The current model for describing the involvement of β-catenin in colon carcinogenesis suggests that it accumulates in the cell in its unphosphorylated form because of mutations in the β-catenin phosphorylation sites or in associating proteins that regulate β-catenin degradation. These excessive amounts of β-catenin then lead to aberrant expression of its target genes, promoting neoplastic transformation (20, 27). It is believed that NH₂-terminal S/T-phosphorylation of β-catenin is necessary and sufficient to initiate ubiquitination-mediated degradation of the protein in an unidirectional process, thereby reducing its cellular level (1). In contradiction to this assumption, the results presented here indicate that phosphorylation and ubiquitination/degradation might be differentially regulated and that phosphorylated β-catenin can accumulate in the cell, resulting in reduced signaling activity. According to our experiments, aspirin-induced S/T phosphorylation of β-catenin was followed by down-regulating the transcription of a β-catenin/TCF-responsive target gene in SW480 without reducing the total amount of β-catenin. This is consistent with our recently published observations, showing that increasing concentrations of aspirin (0.5–10 mm) did not significantly alter the total β-catenin level in SW948, LoVo, HCT116, and SW480 colorectal cancer cell lines, although the β-catenin-signaling activity was markedly reduced (15). We cannot rule out the possibility that this finding is attributable to the experimental procedure used for detection of the total β-catenin amount, which does not differentiate between the membrane-bound, free, or signaling active β-catenin pools. Alternatively, S/T-phosphorylated β-catenin accumulates in SW480 cells without being sufficiently degraded, and this phosphorylation might have a regulatory role in β-catenin signaling in addition to its effect on protein stability. This notion is supported by a recent report demonstrating that S/T-phosphorylated β-catenin forms a transcriptionally inactive complex with lymphoid enhancer-binding factor-1, although it can accumulate in the nucleus (28). Furthermore, it was shown that only unphosphorylated but not phosphorylated β-catenin is able to activate transcription from the TCF-responsive reporter (pTopflash; Ref. 29), which was also used in our experiments.

As outlined above, aspirin did not display any appreciable impact on β-catenin phosphorylation in 293 cells, which correlated well with the moderate effect of the NSAID on down-regulation of β-catenin/TCF signaling in these cells. It is well accepted that the total amount of β-catenin is much lower in 293 cells compared with colorectal cancer cells (28), and effects might therefore not be visible. Interestingly, the total amount of β-catenin in 293 cells was reduced upon aspirin treatment when the cells were grown under serum deprivation, although phosphorylation of the protein was not detectable. Our finding supports the observations described by Staal et al. (29) that elevation or reduction of the total amount of β-catenin is not sufficient to modulate the signaling activity. Taking together the data from SW480, SW948, and 293 cells, it appears likely that S/T-phosphorylation might directly affect the transactivation capacity of β-catenin in cultured cells and that induction of NH₂-terminal β-catenin phosphorylation is the cause of aspirin- and indomethacin-mediated reduction of β-catenin/TCF transcriptional activity.

In principle, an increased level of phosphoralyted β-catenin can be achieved by three distinct mechanisms: (a) by an increased activity of its upstream kinases (CKIα and GSK-3β); (b) by specific inhibition of the interaction between the ubiquitination/degradation machinery and phosphorylated β-catenin; or (c) by inhibition of a phosphatase, responsible for reverting β-catenin to its unphosphorylated form. Several findings argue against the first possibility assuming that the kinase(s) responsible for S/T-phosphorylation of β-catenin may be direct targets of aspirin action. We could demonstrate in our study that the pattern of GSK-3β activity (indicated by phosphorylation at residue S9) in response to aspirin was contradictory to the phosphorylation pattern of β-catenin, namely GSK-3β was not activated by aspirin in SW480 cells despite the increase of phosphorylated β-catenin. Quite the reverse, the enzyme was also phosphorylated in response to aspirin, indicating its inactivation in several colorectal cancer cell lines. Furthermore, although S9-GSK3-β phosphorylation in colorectal cancer cells continuously increased upon 24 h of aspirin treatment, it was reduced in 293 cells after a transient increase (inactivation) of 30 min. The reason for this cell type-specific response is currently unclear. However, if aspirin would bind directly to GSK-3β to affect its activity, one should expect the same response in different cell types, and the response should be visible much earlier. Therefore, direct interaction of aspirin with GSK-3β seems not to be the cause of NSAID-induced NH₂-terminal β-catenin phosphorylation in colon cancer cells. Rather, this points to an indirect mechanism whereby GSK-3β and β-catenin are both affected by a third enzyme, the activity of which is modified by aspirin. Nevertheless, our finding that the aspirin- and indomethacin-induced phosphorylation of β-catenin could be diminished by the GSK-3β inhibitor LiCl indicates participation of GSK-3β in that process, although it is not the only cause of the increased phospho-β-catenin level. Currently, we are unable to answer the question whether the priming kinase CKIα, phosphorylating β-catenin at serine 45, is affected by aspirin or indomethacin, and the precise mechanism by which GSK-3β is targeted by aspirin action remains to be determined.

The interaction between the ubiquitination/degradation machinery and phosphorylated β-catenin is mediated by a mechanism shared with several cellular proteins, i.e., IκBα, β, and ε. The phosphorylated NH₂-terminal recognition motif DSGψXS is recognized by F-box proteins that serve as intracellular receptors for phosphorylated β-catenin or IκB to ensure their prompt and efficient proteolysis (30–32). If aspirin interfered with this recognition machinery, one would expect that the phosphorylation/degradation pattern of IκB should be the same like that of β-catenin in response to aspirin treatment. However, our experiments show that aspirin exerts different effects on phosphorylation of the recognition sites and the subsequent degradation of the corresponding molecules. Therefore, it seems unlikely that aspirin targets F-box proteins or other members of the recognition/degradation complexes.

Our results, demonstrating that GSK-3β phosphorylation at serine 9 is also elevated by aspirin in colon cancer cells, suggest that the NSAID might stabilize selected S/T-phosphorylations by inhibiting a phosphatase, which usually reverts the proteins into their unphosphorylated form. In contrast to the type 1 and 2A serine phosphatase inhibitor calyculin A, which irreversibly stabilizes any S/T-phosphorylation in any cell type within a few minutes (our own data and Ref. 33), we demonstrate that aspirin selectively stabilizes the S/T-phosphorylation of β-catenin and GSK-3β but not that of IκB in a cell type-specific manner. In 293 cells, IκB was even increasingly dephosphorylated in response to aspirin. In addition, aspirin stabilized the phosphorylation of protein kinase B (PKB/Akt) at T308 but not at amino acid S473 in SW480 and SW948 cells, whereas calyculin A increased the phosphorylation level at both regulatory PKB/Akt sites (S. D., unpublished data), whereby the aspirin-induced phosphorylation pattern of PKB/Akt and its time course were exactly the same like that of β-catenin and GSK-3β. Also in contrast to calyculin A, as pointed above, aspirin-induced stabilization of the phosphorylation is not visible before 6–24 h of incubation with the NSAID, indicating that aspirin might indirectly lead to inactivation of a specific phosphatase targeting GSK-3β and β-catenin. The precise nature of this phosphatase, however, remains to be determined.

In summary, the experimental data presented here show that aspirin and indomethacin stabilize phosphorylation of a growth stimulatory protein, β-catenin. This phosphorylation is predominantly responsible for a reduced β-catenin/TCF-signaling activity in colorectal cancer cells without significantly affecting the total amount of β-catenin. The phosphorylation response seems to be cell type specific because it was evident in several colorectal but not in noncolorectal cancer cells and is likely mediated by inhibition of a specific S/T-phosphatase. Although it is controversially discussed whether similar drug concentrations as those used in our and others’ in vitro experiments might be achieved in serum and synovial fluid of patients (16, 34), these data underscore the relevance of β-catenin as a key target for anticancer therapies and the correlation of antitumor effects with the underlying molecular mechanisms will help to develop new chemopreventive agents.