APC Is Essential for Targeting Phosphorylated β-Catenin to the SCFβ-TrCP Ubiquitin Ligase

Molecular Cell

Article APC Is Essential for Targeting Phosphorylated b-Catenin to the SCFb-TrCP Ubiquitin Ligase YunYun Su,1,5 Chunjiang Fu,1,5 Shinji Ishikawa,1,5,6 Alessandra Stella,1,5,8 Masayuki Kojima,1,5,7 Kazuhisa Shitoh,1,5,7 Emanuel M. Schreiber,2 Billy W. Day,2,3,4 and Bo Liu1,5,* 1Department

of Pathology and Proteomics Core Labs 3Department of Pharmaceutical Sciences 4Department of Chemistry 5Hillman Cancer Center University of Pittsburgh, Pittsburgh, PA 15213, USA 6Present address: Department of Surgery II, Kumamoto University Medical School, Kumamoto 860, Japan 7Present address: Department of Surgery, Jichi Medical School, Shimotsuke-shi, Tochigi-ken 329-0498, Japan 8Present address: Department of Human Genetics, Bari University School of Medicine, Bari, Italy *Correspondence: [email protected] DOI 10.1016/j.molcel.2008.10.023 2Genomics

SUMMARY

Ubiquitin-dependent proteolysis is an important mechanism that suppresses the b-catenin transcription factor in cells without Wnt stimulation. A critical step in this regulatory pathway is to create a SCFb-TrCP E3 ubiquitin ligase binding site for b-catenin. Here we show that the SCFb-TrCP binding site created by phosphorylation of b-catenin is highly vulnerable to protein phosphatase 2A (PP2A) and must be protected by the adenomatous polyposis coli (APC) tumor suppressor protein. Specifically, phosphorylated b-catenin associated with the wild-type APC protein is recruited to the SCFb-TrCP complex, ubiquitin conjugated, and degraded. A mutation in APC that deprives this protective function exposes the N-terminal phosphorylated serine/threonine residues of b-catenin to PP2A. Dephosphorylation at these residues by PP2A eliminates the SCFb-TrCP recognition site and blocks b-catenin ubiquitin conjugation. Thus, by acting to protect the E3 ligase binding site, APC ensures the ubiquitin conjugation of phosphorylated b-catenin. INTRODUCTION b-catenin is a transcription factor activated by Wnt signal stimulation during animal development and tissue homeostasis (reviewed by Logan and Nusse, 2004; Clevers, 2006). Although this activation is essential for stimulated cells, it is equally important for unstimulated cells to have a mechanism in place that can effectively prevent an abnormal activation of b-catenin. Ubiquitin-dependent proteolysis is an important mechanism that inactivates b-catenin in cells without Wnt stimulation. This regulatory pathway begins with the phosphorylation of b-catenin at its N-terminal conserved serine residues 33, 37, and 45, and threonine residue 41 (S33/S37/T41/S45). Phosphorylation of b-catenin occurs in a multiprotein complex consisting of adeno-

matous polyposis coli (APC) tumor suppressor protein, Axin, glycogen synthase kinase-3b (GSK-3b), and casein kinase 1 (CK1), among others (Behrens et al., 1998; Hart et al., 1998; Nakamura et al., 1998). Axin acts as a scaffold protein and contains distinctive binding domains for APC, b-catenin, GSK-3b, and CK-1. The role of APC in this complex is less clear, but it is generally believed that interactions among these proteins facilitate the phosphorylation of b-catenin by CK-1 and GSK-3b through a dual kinase mechanism (Liu et al., 2002). Phosphorylation of S33 and S37 creates a consensus b-TrCP recognition site at the N-terminal domain of b-catenin. After being released from the kinase complex, phosphorylated b-catenin is recognized by bTrCP and recruited to the Skp1/Cul1/F-boxb-TrCP (SCFb-TrCP) E3 ubiquitin ligase (Marikawa and Elinson, 1998; Kitagawa et al., 1999; Hart et al., 1999). Ubiquitin-conjugated b-catenin is subsequently degraded by the 26S proteasome. Such a degradation pathway is essential for the normal development as well as for the tumor suppression (Clevers, 2006). Abnormal stabilization of bcatenin has been found to cause several types of human cancer. Mutation in APC is perhaps the most common mechanism that stabilizes b-catenin during oncogenic transformation and the development of colorectal cancer (CRC) (reviewed by Polakis, 1997). APC mutation was originally identified as a genetic cause of familial adenomatous polyposis (FAP) (reviewed by Vogelstein and Kinzler, 2002). The first evidence linking the APC tumor suppressor to b-catenin came to the light when the two proteins were found to interact (Su et al., 1993; Rubinfeld et al., 1993). Subsequent studies reveal that one of the hallmarks of APC mutation is the stabilization and high levels accumulation of b-catenin in mutant cells (Clevers, 2006; Polakis, 2007). Stabilized b-catenin then enters into the nucleus and activates Wnt target genes. Introduction of full length, wild-type (WT) APC to the APC mutant cells reduces the high levels of b-catenin (Munemitsu et al., 1995) and suppresses the transcriptional activation of Wnt target genes, such as MYC (c-myc) and CCND1 (cyclin D1; He et al., 1998; Tetsu and McCormick, 1999). APC has been suggested to regulate b-catenin in a number of ways. First, it may promote the export of nuclear b-catenin upon the withdrawal of Wnt signaling (Henderson, 2000; Neufeld et al.,

652 Molecular Cell 32, 652–661, December 5, 2008 ª2008 Elsevier Inc.

Molecular Cell Regulation of b-Catenin by APC

Figure 1. APC Restores b-Catenin Ubiquitination in CRC Tumor Cells (A) Cell extracts from DLD1, HT29, and SW480 were incubated with (+) or without ( ) myc-tagged wild-type APC (myc-APC) in the presence (+) or absence ( ) of proteasome inhibitor ALLN. Following incubation for 4 hr at room temperature, b-catenin was immunoprecipitated (IP) by a rabbit antib-catenin antibody. Levels of ubiquitin-conjugated and nonconjugated b-catenin in the IP products were determined by immunoblotting (IB) with a mouse anti-b-catenin (upper panel), a rabbit anti-phospho-S33/S37/T41 b-catenin-specific antibody (second panel), and a mouse anti-ubiquitin antibody (third panel), respectively. The bottom panel shows that equal amount of myc-APC was used in relevant reactions. P-bcatenin designates phosphorylated b-catenin. (B) Phosphorylation of b-catenin was reconstituted in vitro as described in the Experimental Procedures. Components included in the reaction were designated on the top of each lane. In all reactions, b-catenin was used as a GST-fusion protein. Following the incubation, levels of Flag-tagged phospho-b-catenin (Flag-P-b-catenin) were determined by IB with a rabbit anti-phospho-S33/S37/T41 b-catenin specific antibody (upper panel). The lower panel shows that equal amount of input Flag-b-catenin was used in each reaction.

2000). Second, it may sequester b-catenin in the cytoplasm, thereby preventing its association with Tcf4 in the nucleus (Neufeld et al., 2000; Rosin-Arbesfeld et al., 2000). The last and perhaps the most important function of APC is linked to b-catenin degradation (Polakis, 2007; Clevers, 2006). Paradoxically, this later function can be substituted by overexpression of Axin or b-TrCP in APC mutant CRC cells (Nakamura et al., 1998; Behrens et al., 1998; Hart et al., 1998, 1999). APC is a large protein with an estimated molecular weight of 330 kDa and contains two separated domains capable of interacting with b-catenin. The first domain consists of three 15 amino acid (aa) repeats, whereas the second domain contains seven 20 aa repeats (Polakis, 1997). Mutational analyses of human CRC indicate that mutant APC proteins often retain the first domain but have truncations wherein parts of or the entire second domain are missing (Vogelstein and Kinzler, 2002). This second domain also contains three 25 aa SAMP (Ser-Ala-Met-Pro) repeats. The SAMP repeats are dispersed among the 20 aa repeats and interact with Axin. Biochemical and genetic studies have shown that both 20 aa and SAMP repeats are required for the downregulation of b-catenin. It is thus suggested that APC may act to promote b-catenin to interact with Axin in the phosphorylation complex like a scaffold protein (Bienz and Clevers, 2000). Yet it is not clear why two scaffold proteins should be required in the same phosphorylation complex. In addition, Axin already contains specific interaction domains for the kinases and substrates in the complex. Unlike the classical scaffold protein Axin, APC does not contain a kinase docking site and is incapable of interacting with GSK-3b and CK-1. Accordingly, phosphorylation of b-catenin has been shown to occur in APC mutant tumor cell lines and in a kinase reaction without the presence of the WT APC (Liu et al., 2002; Xing et al., 2004; Ha et al., 2004; Yang et al., 2006). Hence, a second model has recently been proposed (Xing et al., 2003, 2004). This model suggests that phosphorylated b-catenin is turned over to

APC from Axin and then released from the kinase complex. However, it is not clear why phosphorylated b-catenin has to be turned over to APC before being released from the kinase complex. Thus, the precise function of APC in association with b-catenin degradation remains to be resolved. In this report, we present evidence demonstrating that APC acts to ensure the ubiquitin conjugation of phosphorylated b-catenin. We show that phosphorylated b-catenin associated with the WT APC protein is recruited to the SCFb-TrCP E3 ubiquitin ligase, conjugated with ubiquitin, and degraded. In the absence of the WT APC protein, phosphorylated b-catenin is rapidly dephosphorylated by serine/threonine protein phosphatase 2A (PP2A). Dephosphorylation by PP2A instantly eliminates the b-TrCP binding site and consequently prevents b-catenin entering into the downstream ubiquitination machinery. These findings provide insight into the functional mechanism of APC and may offer a basis for developing treatment of the disease through the prevention of b-catenin dephosphorylation. RESULTS APC Restores b-Catenin Ubiquitination in APC Mutant Cells To gain a better understanding of APC, we used a reconstituted cellfree system to investigate how WT APC restores the b-catenin degradation in APC mutant cells. A similar system was previously used to explore the biochemical process of b-catenin degradation in Xenopus egg extracts (Salic et al., 2000). We examined a number of cell lines and decided to use extracts prepared from CRC cell lines DLD1, HT29, and SW480. These cell lines are well characterized and known to have mutant APC incapable of promoting b-catenin degradation. Accordingly, they contain very high levels of b-catenin (Morin et al., 1997). We confirmed the existence of high levels of b-catenin in these extracts by immunoblotting (IB) (Figure 1A,

Molecular Cell 32, 652–661, December 5, 2008 ª2008 Elsevier Inc. 653

Molecular Cell Regulation of b-Catenin by APC

epitopes, respectively. Levels of phosphorylated b-catenin generated from the kinase reaction were determined by IB with a phospho-S33/S37/T41 b-catenin-specific antibody (Figure 1B). In the presence of GSK-3b, a basal level of phosphorylation of b-catenin at residues S33/S37/T41 was detected (lane 3). Phosphorylation at these sites was significantly enhanced by CK-1 (lane 4). This was further stimulated by Axin (lanes 6 and 7). However, addition of APC to the reconstituted reaction seemed to have little effect on b-catenin phosphorylation (lanes 5 and 7). The lack of stimulation by APC was reconfirmed using a higher concentration (5-fold) of APC protein (lane 8). The biochemical reconstitution experiments thus confirm the result of cell-free system, suggesting that APC does not have a function in stimulating b-catenin phosphorylation.

Figure 2. Ubiquitin Conjugation of Phosphorylated b-Catenin Requires WT APC (A) DLD1 extracts with (+) or without ( ) myc-APC was immunoprecipitated by a control mouse IgG (lane 1), a mouse anti-b-TrCP (lanes 2 and 3), or a rabbit anti-b-catenin antibody (lanes 4 and 5). Levels of b-catenin (upper panel), b-TrCP (second panel), and Skp1 (bottom panel) in the IP complex were determined by IB with respective antibodies. (B) Flag-b-catenin was phosphorylated in the presence (+) or absence ( ) of myc-APC in vitro. After phosphorylation, free and myc-APC-associated phospho-b-catenin were separated from the rest kinase components and incubated in DLD1, SW480, and HT29 extracts. After incubation for 4 hr at room temperature in the presence (+) or absence ( ) of ALLN, levels of ubiquitinconjugated and nonconjugated b-catenin were determined by IB with a Flag antibody. IB results from DLD1, SW480, and HT29 are shown in the first, second, and third panels, respectively. The forth and fifth panels show amount of input myc-APC and Flag-P-b-catenin.

upper panel, lanes 1, 5, and 9). Further analysis with an antiphospho-b-catenin-specific antibody revealed that a significant amount of b-catenin was phosphorylated (second panel). This supports a previous notion suggesting that APC mutation does not affect b-catenin phosphorylation (Yang et al., 2006). Further examination by treating the extracts with proteasome inhibitor N-acetyl-leucinyl-leucinyl-norleucinyl-H (ALLN) showed that phosphorylated b-catenin was not ubiquitin-conjugated in these extracts (lanes 2, 6, and 10). Ubiquitination of b-catenin was restored by addition of myc-tagged WT APC protein (lanes 4, 8, and 12). Accordingly, in the absence of ALLN, ubiquitinconjugated b-catenin was degraded (lanes 3, 7, and 11). To verify these results, we reconstituted the kinase reaction in vitro using purified protein components, including Axin, APC, GSK-3b, CK-1a, and b-catenin. To facilitate the biochemical analysis of these proteins, the C terminus of Axin, APC, and b-catenin were tagged with hemagglutinin (HA), myc, and Flag

Ubiquitin Conjugation of Phosphorylated b-Catenin Requires APC The above biochemical reconstitution results demonstrated that APC mutation specifically interrupted the ubiquitin conjugation of phosphorylated b-catenin. To explore how b-catenin ubiquitination is disrupted in APC mutant cells, we examined the interaction between phosphorylated b-catenin and SCFb-TrCP in DLD1 extracts by coimmunoprecipitation (IP). IB analysis of a b-TrCP antibody-precipitated protein products showed that b-catenin was not coimmunoprecipitated with SCFb-TrCP from DLD1 extracts (Figure 2A, lane 3), but their interaction was restored by WT APC (lane 2). A reciprocal IP analysis with a b-catenin antibody further confirmed this finding (lanes 4 and 5). To provide additional evidence, we produced WT APC-associated and free phospho-b-catenin from in vitro kinase reactions. Following the separation from the other kinase components, the two forms of phospho-b-catenin were tested in the cellfree system. The result demonstrated that ubiquitin conjugation and the subsequent degradation of b-catenin could only occur to the WT APC-associated (Figure 2B, top panel, lanes 1 and 2) but not the free phospho-b-catenin (lanes 3 and 4). The result was reproducible in SW480 and HT29 extracts (Figure 2B, second and third panels). These findings support a hypothesis that WT APC is essential for targeting phosphorylated b-catenin to the downstream SCFb-TrCP E3 ubiquitin ligase. Phosphorylated b-Catenin Not Associated with WT APC Is Dephosphorylated Our theory that APC is required for the recruitment of phosphorylated b-catenin to the SCFb-TrCP ubiquitin ligase seems to be at odds with a current belief. Indeed, when assessed by in vitro reconstitution using purified protein components, both APCassociated and free phospho-b-catenin interacted at a similar level with the in vitro translated and 35S-labled b-TrCP protein, shown by the co-IP analysis (Figure 3A, upper panel, lanes 2 and 4). Surprisingly, this interaction was abolished if free phospho-b-catenin had been preincubated in DLD1 extracts (lane 5). By contrast, the same preincubation did not affect the interaction of APC-associated phospho-b-catenin (lane 3). It should be noted that myc-APC was also detected in the coimmunoprecipitated protein products (second panel, lanes 2 and 3), suggesting that phosphorylated b-catenin remained associated with WT APC even after it was recruited to the E3 ubiquitin ligase.

654 Molecular Cell 32, 652–661, December 5, 2008 ª2008 Elsevier Inc.

Molecular Cell Regulation of b-Catenin by APC

phospho-b-catenin produced in the absence of WT APC is rapidly dephosphorylated.

Figure 3. Phosphorylated b-Catenin Not Associated with WT APC Is Rapidly Dephosphorylated (A) myc-APC associated (lanes 1–3) or free phospho-b-catenin (lanes 4 and 5) was incubated with the TNT (in vitro transcription-coupled translation) reaction mix containing a mock translated (lane 1) or 35S-labeled b-TrCP protein (lanes 2 to 5). Samples in lanes 3 and 5 were preincubated in DLD1 extracts for 10 min prior to the incubation with 35S-labeled b-TrCP. Following the incubation, proteins associated with 35S-labeled b-TrCP were immunoprecipitated with a mouse anti-b-TrCP antibody. Levels of Flag-phospho-b-catenin in the IP products were determined by IB with a rabbit anti-phospho-S33/S37/T41 b-catenin-specific antibody (upper panel). Amount of myc-APC in the same IP products was determined by IB with a myc-antibody (second panel). The third panel is an autoradiography of 35S-labeled b-TrCP from the IP complex. The forth panel is an autoradiography of input 35S-labeled b-TrCP from TNT lysates. The bottom panel indicates that equal amount of Flag-phospho-b-catenin was used in each reaction. (B) Free (left) and WT APC-associated Flag-[32P]-labeled phospho-b-catenin (right) were incubated in DLD1, SW480, and HT29 extracts for 0, 10, 20, and 30 min (min) at room temperature as indicated. After the incubation, levels of radiation signal representing Flag-[32P]-b-catenin from DLD1, SW480, and HT29 were determined by autoradiography and are shown from top to the third panels as designated. The bottom panel shows that equal amount of input Flag-[32P]-b-catenin used in each reaction.

The above results suggested that dephosphorylation might have occurred to the free phospho-b-catenin during its preincubation in DLD1 extracts. To validate this possibility, we generated WT APC-bound and free [32P]-labeled phospho-b-catenin from in vitro kinase reaction. Upon the separation from other kinase components, the [32P]-labeled phospho-b-catenin was chased in the DLD1 extracts at different time intervals of 10, 20, and 30 min. Direct autoradiography revealed that [32P]-labeled radiation signals disappeared rapidly from the free phosphob-catenin shortly after 10 min of incubation (Figure 3B, upper panel at left), whereas such a signal loss was less obvious in the WT APC-associated [32P]-phospho-b-catenin (upper panel at right). Further testing in extracts prepared from SW480 and HT29 cells demonstrated a similar result (Figure 3B, second and third panels). These findings, taken together, show that

Protein Phosphatase 2A Dephosphorylates Phospho-b-Catenin Bovine cardiac muscle (BCM) is known to contain very high phosphatase activity (Mumby et al., 1987). Consistent with this, crude extracts of BCM caused a rapid dephosphorylation of free phospho-b-catenin. This dephosphorylation was effectively blocked by a protein phosphatase (PP) inhibitor okadaic acid (OA) (data not shown). We thus attempted to identify this phospho-b-catenin-specific phosphatase from the BCM extracts. Biochemical purification results showed that this phospho-b-catenin-specific phosphatase consisted of two different subunits with molecular weight of approximately 63 and 38 kDa when examined by SDS-PAGE (Figure 4A). C18-LC-nanoESI-ion trap MSn mass spectrometric analysis of the trypsindigested 63 kDa subunit gave results of high confidence that the protein was PP2A Aa (Figure 4B). Consistent with this result, the 63 kDa subunit was specifically recognized by an antibody raised against the PP2A Aa subunit, whereas the 38 kDa subunit was recognized by an antibody raised against the PP2A catalytic subunit Ca (PP2A Ca) (Figure 4C). The protein purification and mass spectrometric analysis indicated that the PP2A core enzyme was responsible for the rapid dephosphorylation of free phospho-b-catenin. Consistent with this conclusion, the dephosphorylation in the DLD1 extracts was effectively inactivated by OA (Figure 4D, top panel, lane 1) at a concentration that was known to be effective on PP2A but not on most of other known phosphatases (reviewed by Janssens and Goris, 2001). To provide additional evidence, we absorbed DLD1 extracts onto PP2ACa antibody-conjugated affinity beads. This treatment removed greater than 80% of PP2A based on immunoblot analysis with an anti-PP2A Ca antibody (second panel). Accordingly, dephosphorylation of b-catenin was dramatically reduced in PP2A-deficient extracts (top panel, lane 3). In contrast, immunodepletion of the other known PPs, such as PP1 (lane 4), PP2B (lane 5), PP2C (lane 6), and PP5 (lane 7), did not have such an obvious effect. Deficiency in PP2A Leads to an Excessive Degradation of Membrane-Associated b-Catenin To further test our findings in vivo, small interfering RNA (siRNA) specific to the PP2A Ca catalytic subunit was introduced into cultured DLD1 cells by the transient transfection. A dramatic reduction in b-catenin was observed (Figure 5A, upper panel) following siRNA-mediated reduction of PP2A Ca (bottom panel). Similar levels of reduction in b-catenin were also evident in HT29 and SW480 cells treated with the siRNA (data not shown). Noticeably, following the degradation of cytoplasmic b-catenin in DLD cells (compare lanes 3 and 9), a reduction in membrane E-cadherin-associated b-catenin also occurred (compare lanes 5 and 11). The reduction of both cytosolic and E-cadherin-associated b-catenin was the result of ubiquitination, evidenced by accumulation of b-catenin-ubiquitin conjugates in the presence of ALLN (lanes 10 and 12). Most surprisingly, this degradation seemed to include the entire membrane E-cadherin/b-catenin adhesion protein complex, despite the fact that E-cadherin

Molecular Cell 32, 652–661, December 5, 2008 ª2008 Elsevier Inc. 655

Molecular Cell Regulation of b-Catenin by APC

Figure 4. PP2A Is Phospho-b-Catenin-Specific Phosphatase (A) Phospho-b-catenin-specific phosphatase was purified near homogeneity from crude extracts of bovine cardiac muscle. Two micrograms purified phosphatase was resolved by 12% SDS-PAGE. Coomassie blue staining of the gel revealed two major protein subunits. (B) Mass spectral (C18-LC-nanoESI-ion trap MSn) analysis of the trypsin-digested 63 kDa subunit. Peptides with high confidence matched to the data from the NCBI database are shown in their order of elution from the C18 column. Data in the columns are as follows. Scan number, MS scan used; sequence, predicted amino acid sequence of the tryptic peptide (an asterisk after a methionine (M*) indicates mono-oxygenation of the thioether sulfur); theoretical [M+H]+, the predicted mass-to-charge ratio for the singly charged, protonated molecular ion; peptide charge state, the actual charge state of peptide ion; crosscorrelation, values from the search; delta correlation, describes how different the first hit is from the second hit in the search results; preliminary score, the higher the value the better; rank/preliminary score, the rank of the match during the preliminary scoring; and MS/MS ions, the number of the experimental MS/MS ions matched with the theoretical MS/MS ions for the listed peptide. (C) 100 ng of above-purified protein was resolved by 12% SDS-PAGE and probed by a PP2A Aa and Ca antibodies. (D) DLD1 cytosolic extracts were treated with 10 nM okadaic acid (OA) (top panel, lane 1) or affinity beads conjugated with a control mouse IgG (lane 2) and antibodies against the respective phosphatases (designated on the top). Levels of each phosphatase in the antibody-depleted extracts were determined by IB with their respective antibodies and are shown from the second to sixth panels. Lysates confirmed to have greater than 80% of reduction in phosphatase levels were used to incubate with Flag-[32P] phospho-b-catenin for 10 min at room temperature. After incubation, levels of Flag-[32P] phospho-b-catenin were determined by direct autoradiography (upper panel). The bottom panel indicates that equal amount of Flag-[32P] phospho-b-catenin was used in each reaction.

was not ubiquitinated (Figure 5A, second panels; compare lanes 5 and 6 with lanes 11 and 12). Such a degradation in the membrane-associated E-cadherin/b-catenin protein complex also occurred in a control cell line 293T treated with the same PP2A-specific siRNA (Figure 5B). The excessive degradation in membrane-associated b-catenin could not be rescued even when these siRNA-treated cells were grown on a Wnt-containing medium (data not shown). APC Protects the b-TrCP Binding Site from PP2A-Mediated Destruction If APC indeed has a function in protecting the b-TrCP recognition site of phosphorylated b-catenin, then it should specifically prevent it from PP2A-mediated dephosphorylation. To evaluate this notion, free and APC-associated phospho-b-catenin were incubated with purified bovine PP2A core enzyme for 10, 20, and 30 min. IB with a phospho-b-catenin-specific antibody showed that while free phospho-b-catenin was rapidly dephosphory-

lated by PP2A (Figure 6A, upper panels), phospho-b-catenin associated with APC remained mostly unaffected (second panels). To test the specificity of this function, we compared APC with two other known phospho-b-catenin binding proteins, Axin and b-TrCP. The result of this test showed that phosphorylated b-catenin bound by b-TrCP was resistant to PP2A-mediated dephosphorylation, albeit at a significantly reduced level compared with that of APC (third panels). In contrast, phosphorylated b-catenin associated with HA-Axin was rapidly dephosphorylated (fourth panels). The above assays demonstrated that APC specifically protected the b-TrCP recognition site from PP2A-mediated destruction. If this protection is essential for APC to suppress b-catenin, then mutant APC should be defective in this activity. To validate this possibility, we analyzed three mutant APC proteins commonly found in CRC tumors. These were APC1 from SW480, APC2 from DLD1, and APC3 from HT29 (mutant and WT APC protein structures are illustrated in Figure 6B). Following the

656 Molecular Cell 32, 652–661, December 5, 2008 ª2008 Elsevier Inc.

Molecular Cell Regulation of b-Catenin by APC

Figure 5. Deficiency in PP2A Causes an Excessive Degradation of Membrane-Associated b-Catenin (A) DLD1 cells were transiently transfected with either the control GFP or the PP2A Ca-specific siRNA. After 72 hr, cells were fractioned and prepared as whole cell (W), cytosolic (C), and membrane protein lysates (M). For detection of ubiquitin conjugates, ALLN was added into the culture medium 6 hr before the harvest. Levels of b-catenin, E-cadherin, and PP2A Ca in each fraction are displayed from top to bottom panels. IB results from control siRNA transfected cells are shown on the left; those transfected with PP2A Ca-specific siRNA are shown on the right. (B) IB analysis of 293T cell lysates transfected with the above siRNA. Cell fractions and IB antibodies were essentially the same as in (A). Levels of b-catenin, E-cadherin, and PP2A Ca are displayed from top to bottom panels.

in vitro kinase reaction, mutant APC-associated phospho-b-catenin was incubated with purified bovine PP2A core enzyme. Analysis by IB showed that phospho-b-catenin in the mutant APC-associated complexes was rapidly dephosphorylated by PP2A (Figure 6C). Noticeably, unlike the WT APC protein, which dramatically increases in its binding affinity with b-catenin following the kinase reaction, the binding affinity of these mutant APC proteins remained mostly unchanged before or after the kinase reaction (data not shown). Previous studies have shown that a fragment of APC containing both 20 aa and SAMP repeats is sufficient to suppress b-catenin (Shih et al., 2000). To determine whether the abovedefined APC function may coincide within this fragment (herein designated as APC4, encoded by codons 1265–2075), we cloned and expressed APC4. The purified APC4 protein was then evaluated for its ability to protect the b-TrCP binding site of phosphob-catenin. The result confirmed that this domain contained sufficient activity to protect the integrity of the b-TrCP binding site from PP2A-mediated destruction (Figure 6C). Phosphorylation of APC Is Required for Its Protective Function The 20 aa repeats of APC contain multiple conserved serine residues that are the phosphorylation targets of CK-1 and GSK-3b (Bhattacharya and Boman, 1995; Rubinfeld et al., 2001). Phosphorylation in these 20 aa repeats has been shown to dramati-

cally increase the binding affinity of APC (Ha et al., 2004; Xing et al., 2004). To investigate whether phosphorylation of APC is required for protecting the b-TrCP binding site, we prepared phospho- and nonphospho-APC4-associated phospho-b-catenin. The nonphospho-APC4-associated protein complex was generated by combining phosphorylated b-catenin with the nonphosphorylated APC4, whereas the phospho-APC4-associated was generated by including the two proteins in the same kinase reaction. When incubated with purified bovine PP2A core enzyme, phosphorylated APC4 protected the b-TrCP binding site, whereas nonphosphorylated APC4 did not (Figure 7A). These data confirm a previous hypothesis suggesting that phosphorylation is important for APC function (Rubinfeld et al., 2001; Ha et al., 2004; Xing et al., 2004). It was evident that following the kinase reaction, the binding affinity between APC4 and b-catenin was dramatically increased. This tight association could not be disrupted by a GST-fusion protein containing the b-catenin binding domain of Axin (GST-bBD) or the 15 aa repeats of APC (GST-15 aa; Figure 7B). In contrast, nonphosphorylated APC4 had much weaker binding with b-catenin, which could be effectively competed off by the GST-bBD and the GST-15 aa fusion proteins. It was noticeable that the Axin bBD domain appeared to have significantly higher affinity to b-catenin than that of APC-15 aa repeats. These results are consistent with previous biochemical studies indicating that phosphorylation dramatically increases APC binding to b-catenin (Rubinfeld et al., 2001; Ha et al., 2004; Xing et al., 2004). DISCUSSION On the bases of these findings, we propose that a critical function of APC is to provide an immediate protection for the SCFb-TrCP binding site created by phosphorylation of b-catenin. This activity ensures that b-catenin will be targeted to its downstream SCFb-TrCP ubiquitination ligase, conjugated with ubiquitin, and degraded. The physiological importance of this protective function is best illustrated in CRC tumor cells as well as in APC mutant animal models containing high levels of b-catenin. A mutation in APC that deprives this protective function exposes the N-terminal phosphorylated serine/threonine residues of b-catenin to PP2A in APC mutant cells. While b-catenin is continuously phosphorylated by kinases, phosphorylated b-catenin released from the kinase complex is instantaneously reversed by PP2A. Dephosphorylation by PP2A destroys the SCFb-TrCP binding site created by phosphorylation. As a result, b-catenin is prevented from been recruited to the SCFb-TrCP complex for ubiquitin conjugation, leading to its stabilization and high levels of accumulation. This model supports a previously proposed hypothesis that phosphorylated b-catenin must be turned over to APC before it is released from the kinase complex (Xing et al., 2003, 2004). It is conceivable that a failure in turning over phosphorylated b-catenin to APC will result in an immediate reversal of phosphorylation, leading to the stabilization of the b-catenin protein. Likewise, it offers an explanation for a long-standing paradox of why overexpression of Axin or b-TrCP can reduce high levels of b-catenin in APC mutant CRC cells (Nakamura et al., 1998;

Molecular Cell 32, 652–661, December 5, 2008 ª2008 Elsevier Inc. 657

Molecular Cell Regulation of b-Catenin by APC

Figure 6. APC Protects the b-TrCP Binding Site of Phospho-b-Catenin from PP2A-Mediated Destruction (A) 200 nM of phosphorylated b-catenin was incubated with about equal molar of myc-P-APC, HA-Axin, and b-TrCP to form respective protein complex. After 30 min of incubation, each reaction was divided into two separated groups. The control group was supplemented with 25 nM of BSA; the experimental group was mixed with 25 nM of PP2A core enzyme. Aliquots of samples were taken at 0, 10, 20, and 30 min of incubation. Levels of Flag-P-b-catenin from each of these time points were examined by IB with a rabbit anti-phospho-S33/S37/T41 b-catenin-specific antibody. IB results from the control group are shown on the left; those of the PP2A-treated group are shown on the right. Listed samples from top to bottom panels are free flag-P-b-catenin and flag-P-b-catenin preincubated with myc-P-APC, b-TrCP, and HA-Axin. (B) Schematic of APC1, APC2, APC3, APC4, and the WT APC. Protein domains and repeats sequences are designated. (C) Flag-b-catenin was phosphorylated in the presence of WT, APC4, and three mutant APC proteins as designated on the top. Following the phosphorylation, the Flag-P-b-catenin/APC protein complex was separated from other kinase components and incubated with 25 nM PP2A core enzyme for 20 min. After the incubation, levels of phospho-b-catenin from each reaction were examined by IB with a rabbit anti-phospho-S33/S37/T41 b-catenin-specific antibody (upper panel). The second and third panels indicate amount of input Flag-P-b-catenin and APC proteins detected with a flag and a myc antibody.

Behrens et al., 1998; Hart et al., 1998, 1999). Based on our model, it can be reasoned that an elevation of Axin by overexpression may significantly increase the activities of phosphorylation and thus enhance the output of phosphorylated b-catenin. This may sufficiently counter the effect of PP2A and thus reduce the high levels of b-catenin in the absence of WT APC. Conversely, an

elevation of b-TrCP may significantly accelerate the recruitment of phosphorylated b-catenin by the SCFb-TrCP ubiquitination ligase in APC mutant cells, resulting in the reduction of high levels of b-catenin. Preliminary tests in a cell-free system using purified HA-Axin and Flag-b-TrCP proteins seems to support these notions (see the Supplemental Data available online). Figure 7. Phosphorylation Is Required for the Protective Function of APC

(A) Flag-P-b-catenin associated with nonphosphorylated and phosphorylated APC4 (designated as P-APC4) were incubated with purified PP2A core enzyme as described in Figure 6 for 20 min. Following the incubation, levels of Flag-P-bcatenin were determined by IB with a rabbit antiphospho-S33/S37/41 b-catenin-specific antibody (upper panel). The second and third panels show equal amount of input Flag-P-b-catenin and myc-APC4 was used in the reaction. (B) Nonphosphorylated and phosphorylated-mycAPC4/Flag-P-b-catenin (designated as APC4/Pb-cat and P-APC4/P-b-cat, respectively) protein complexes were subjected to binding competition assays by incubation with increasing amounts of GST-15 aa and GST-bBD fusion proteins (shown at molar excesses of 1-, 10-, and 100-fold), respectively. After the incubation, the reaction mix was immunoprecipitated with an anti-myc antibody. Levels of Flag-P-b-catenin that remained associated with nonphosphorylated and phosphorylated mycAPC4 in the IP products were determined by IB with a Flag antibody (upper panel). The lower panel shows amount of myc-APC4 in the IP products detected by an myc antibody.

658 Molecular Cell 32, 652–661, December 5, 2008 ª2008 Elsevier Inc.

Molecular Cell Regulation of b-Catenin by APC

At the present time, we do not know how APC protects the SCFb-TrCP binding site of phospho-b-catenin. It is possible that APC, following its binding with phospho-b-catenin, may create a unique spatial structure, making the N-terminal phosphoserine/threonine residues inaccessible to PP2A. Notably, this protective function does not seem to interfere with the interaction between phospho-b-catenin and b-TrCP. Consistent with this notion, our in vitro binding assays demonstrate that both free and APC-associated phospho-b-catenin interact with in vitro translated b-TrCP without noticeable difference (see Figure 3). It is expected that a final resolution of this structure may require the cocrystallization and structural analysis of a fairly large APC/ phospho-b-catenin complex consisting of multiple domains. Such a study may require the inclusion of an APC fragment encompassing both 20 aa and the SAMP repeats in complex with a phosphorylated b-catenin fragment containing both the armadillo and N-terminal phosphorylation domains. It can be envisioned that future revelation of the hypothesized spatial structure should shed a clear light into the APC protective function. Despite the lack of structural knowledge, it seems that tight binding between the two proteins could be fundamental. Such an interaction seems to be a result of phosphorylation at 20 aa repeats of APC by CK-1 and GSK-3b as shown by the previous studies (Bhattacharya and Boman, 1995; Rubinfeld et al., 2001; Ha et al., 2004; Xing et al., 2004). Consistent with this notion, we found that nonphosphorylated APC had weak binding to phosphorylated b-catenin and could not prevent PP2A-mediated dephosphorylation. Moreover, it is likely that a direct physical interaction with Axin might be essential for APC phosphorylation. For example, the three APC mutants that we have tested here contain only the 15 and 20 aa repeats. However, none of them retains any SAMP repeat required to interact with Axin. Accordingly, these mutants demonstrate only weak binding that is incapable of providing a protective function. Consequently, phosphorylated b-catenin is instantaneously dephosphorylated by PP2A, leading to its high levels of accumulation. Several earlier studies have provided evidence suggesting that PP2A may play an important role in the Wnt signaling pathway (reviewed by Millward et al., 1999; Janssens and Goris, 2001). Although the precise function of PP2A in the Wnt signaling pathway remains to be better defined, our present findings suggest that it has a profound effect on the stability of the b-catenin protein. This notion bodes with the dual functional mechanisms of b-catenin. b-catenin, in addition to serving as a transcription factor for Wnt signaling, is an indispensable component of the adhesion complex (reviewed by Gumbiner, 2000; Jamora and Fuchs, 2002). In unstimulated cells, though the cytoplasmic b-catenin is constantly phosphorylated and degraded, the E-cadherin-associated b-catenin must be stabilized to perform its adhesion function (Yap et al., 1997). The discovery of an instantaneous dephosphorylation of the non-APC-associated phospho-b-catenin by PP2A should have an important physiological ramification for the adhesion-associated b-catenin. For example, in the absence of PP2A, such as those in PP2A knockout mouse embryos, hyperphosphorylation occurs to the membrane-associated b-catenin, resulting in its excessive degrada} tz et al., 2000). More importantly, excess degradation tion (Go of b-catenin in the PP2A knockout embryonic cells appears to

be a dominant event that could not be rescued by Wnt stimula} tz et al., 1998). As a result, PP2A null embryos die at tion (Go a very early stage of development. Consistent with these findings, we show that depletion of PP2A in cultured cells by siRNA dramatically reduced the total levels of b-catenin. While high levels of cytoplasmic b-catenin were dramatically reduced, the membrane-associated b-catenin was also targeted by ubiquitin proteolysis. The result supports an earlier notion suggesting that PP2A might be required to maintain the stability of the membrane } tz et al., 2000). Although E-cadherin/b-catenin complex (Go we did not detect any ubiquitin-conjugated E-cadherin, this protein was nevertheless degraded concurrently with the disappearance of membrane-associated b-catenin. It appears that deficiency in PP2A will result in a total destruction to the entire cellular pools of b-catenin, regardless of its function or location. In contrast, APC-mediated destruction is highly specific and targets only the transcription-related b-catenin. Finally, the fact that phosphorylated b-catenin must be protected by APC to prevent an immediate reversal by PP2A raises an important question: Is the mechanism described here applicable only to the b-catenin transcription factor? As protein phosphatases are ubiquitously and abundantly expressed, it is possible that a similar protective mechanism might be in play in other phosphorylation-regulated cellular processes. This hypothesis can be tested in future studies. EXPERIMENTAL PROCEDURES Phosphorylation of b-Catenin In Vitro The in vitro phosphorylation of b-catenin was performed using the conditions established previously in Xenopus egg extracts with a few modifications (Salic et al., 2000). Briefly, to achieve optimal kinase activities, the components of the reaction were kept at the following concentrations: 10 nM each of GSK-3b and CK-1a, 100 nM of Flag-b-catenin and myc-APC, and 5 nM of HA-Axin. The kinase reactions were then performed in a HEPES-based buffer system (20 mM HEPES [pH 7.6], 10 mM MgCl2, 100 mM NaCl, 1 mM DTT, 1 mM EDTA, 1 mM EGTA, 250 mM ATP, 1 mg/ml BSA, and 0.1% Tween-20). For [32P]-labeling, cold ATP at a 50 mM concentration and 10 mCi of g-[32P]ATP was included in a total reaction volume of 100 ml. To distinguish it from the endogenous b-catenin, GST-Flag-b-catenin (130 kDa) fusion protein was used in the in vitro kinase and the cell-free assay system. To recover free and APC-associated (including both WT and mutant APC) phospho-b-catenin after the kinase reactions, the reaction mixture was absorbed with affinity beads conjugated with anti-HA, anti-GSK-3b, and anti-CK-1a antibodies, respectively. Aliquots of recovered free and APC-associated phospho-b-catenin were examined by IB with HA, GSK-3b, and CK-1a antibodies to confirm that they were free of Axin, GSK-3b, and CK-1a. Construction and purification of recombinant proteins used in our in vitro reconstitution assays are described in the Supplemental Data. Degradation of b-Catenin in a Cell-Free System To establish a cell-free system for the ubiquitin-mediated proteolysis of b-catenin, cytosol was prepared from DLD1, SW480, and HT29 cells using hypotonic buffer (25 mM HEPES [pH 7.6], 5 mM KCl, 2 mM MgCl2, 1 mM DTT, 1% CLAP cocktail, and 1% AEBSF). In vitro ubiquitin conjugation and degradation assays were performed essentially as described previously (Chen et al., 1996), except the reactions were carried out at room temperature. To detect ubiquitin conjugates, 12.5 mM ALLN was used to block the proteasome activities. Protein Purification and Mass Spectrometric Analysis Proteins with phospho-b-catenin-specific phosphatase activity were purified based on the previously described methods (Mumby et al., 1987). Fractions with highest dephosphorylation activity were pooled and concentrated. For

Molecular Cell 32, 652–661, December 5, 2008 ª2008 Elsevier Inc. 659

Molecular Cell Regulation of b-Catenin by APC

mass spectrometric analysis, 10 mg of purified protein was resolved by SDSPAGE. The protein band was cut from the gel and digested with sequencing grade trypsin (Promega) at a 50:1 protein-enzyme ratio; the peptides released from the gel band were analyzed by C18-LC-nanoESI-ion trap MSn in the positive-ion detection mode on a ThermoFinnigan LCQ DECA XP Plus mass spectrometer. Data were analyzed with Sequest and BioWorks software. Immunodepletion Cytosolic protein extracts (1.4–1.6 mg of protein) were first incubated with 10 mg of control IgG (from a combination of mouse and rabbit), mouse antiPP2A catalytic subunit a (clones 46 and 1D6 for both a and b subunits), rabbit anti-PP1, mouse anti-PP2B, anti-PP2C, and anti-PP5/PPT antibodies at 4 C for 2 hr. This was followed by incubation with 20 mg of biotinylated anti-mouse and anti-rabbit ‘‘universal’’ secondary antibody (Vector Laboratories) for 30 min. The antibody-captured protein was then removed by incubation with 20 ml of Streptavidin MagneSphere Paramagnetic Particles (Promega). In most experiments, two rounds of immunodepletion were sufficient to remove greater than 80% of PPs from the cytosolic protein extracts. siRNA-Mediated Knockdown of PP2A Short interfering RNA oligonucleotides (siRNA) for PP2A Ca were synthesized by Ambion. In total, three pairs of oligonucleotides were found to cause obvious reductions in PP2A Ca in a level between 20% and 30%. The rest synthesized oligonucleotides were found less effective and were not used for targeting experiments. The oligo sequences used in transient transfection are listed in the Supplemental Data. Immunoblotting and Immunoprecipitation Immunoblotting (IB) and immunoprecipitation (IP) were performed essentially as described (Harlow and Lane, 1999). The resulting signals were visualized by exposure to X-ray film (Sterling), which were scanned and analyzed for signal strength by Image Quant software (Molecular Dynamics). Antibodies used in our experiments are listed in the Supplemental Data. SUPPLEMENTAL DATA The Supplemental Data include Supplemental Experimental Procedures and one figure and are available online at http://www.molecule.org/supplemental/ S1097-2765(08)00764-8. ACKNOWLEDGMENTS We thank Rebecca Helmer and Xiaoyan Liu for technical assistance and help with cloning and protein purification; Jeff Germuska for obtaining the bovine heart tissues; and Drs. Richard Steinman, Laura Niedernhofer, and Frederick Moolten for helpful discussions and critical reading of the manuscript. This work was supported by National Institutes of Health Grants CA81357 and CA113831, Pennsylvania Health Research grants, the V-Foundation for Cancer Research Award, and the Hillman Scholar Award to B.L.

Chen, Z.J., Parent, L., and Maniatis, T. (1996). Site-specific phosphorylation of IkappaBalpha by a novel ubiquitination-dependent protein kinase activity. Cell 84, 853–862. Clevers, H. (2006). Wnt/b-catenin signaling in development and disease. Cell 127, 469–480. } tz, J., Probst, A., Ehler, E., Hemmings, B., and Kues, W. (1998). Delayed Go embryonic lethality in mice lacking protein phosphatase 2A catalytic subunit Calpha. Proc. Natl. Acad. Sci. U.S.A. 95, 12370–12375. } tz, J., Probst, A., Mistl, C., Nitsch, R.M., and Ehler, E. (2000). Distinct role of Go protein phosphatase 2A subunit Calpha in the regulation of E-cadherin and beta-catenin during development. Mech. Dev. 93, 83–93. Gumbiner, B.M. (2000). Regulation of cadherin adhesive activity. J. Cell Biol. 148, 399–404. Ha, N.C., Tonozuka, T., Stamos, J.L., Choi, H.J., and Weis, W.I. (2004). Mechanism of phosphorylation-dependent binding of APC to beta-catenin and Ats role in beta-catenin degradation. Mol. Cell 15, 511–521. Harlow, E., and Lane, D. (1999). Using Antibodies: A Laboratory Manual (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press). Hart, M.J., de los Santos, R., Albert, I.N., Rubinfeld, B., and Polakis, P. (1998). Downregulation of beta-catenin by human Axin and its association with the APC tumor suppressor, beta-catenin and GSK3 beta. Curr. Biol. 8, 573–581. Hart, M., Concordet, J.P., Lassot, I., Albert, I., del los Santos, R., Durand, H., Perret, C., Rubinfeld, B., Margottin, F., Benarous, R., and Polakis, P. (1999). The F-box protein beta-TrCP associates with phosphorylated beta-catenin and regulates its activity in the cell. Curr. Biol. 9, 207–210. He, T.C., Sparks, A.B., Rago, C., Hermeking, H., Zawel, L., da Costa, L.T., Morin, P.J., Vogelstein, B., and Kinzler, K.W. (1998). Identification of c-MYC as a target of the APC pathway. Science 281, 1509–1512. Henderson, B.R. (2000). Nuclear-cytoplasmic shuttling of APC regulates b-catenin subcellular localization and turnover. Nat. Cell Biol. 2, 653–660. Jamora, C., and Fuchs, E. (2002). Intercellular adhesion, signalling and the cytoskeleton. Nat. Cell Biol. 4, E101–E108. Janssens, V., and Goris, J. (2001). Protein phosphatase 2A: a highly regulated family of serine/threonine phosphatases implicated in cell growth and signaling. Biochem. J. 353, 417–439. Kitagawa, M., Hatakeyama, S., Shirane, M., Matsumoto, M., Ishida, N., Hattori, K., Nakamichi, I., Kikuchi, A., Nakayama, K., and Nakayama, K. (1999). An F-box protein, FWD1, mediates ubiquitin-dependent proteolysis of betacatenin. EMBO J. 18, 2401–2410. Liu, C., Li, Y., Semenov, M., Han, C., Baeg, G.H., Tan, Y., Zhang, Z., Lin, X., and He, X. (2002). Control of beta-catenin phosphorylation/degradation by a dualkinase mechanism. Cell 108, 837–847. Logan, C.Y., and Nusse, R. (2004). The Wnt signaling pathway in development and disease. Annu. Rev. Cell Dev. Biol. 20, 781–810. Marikawa, Y., and Elinson, R.P. (1998). beta-TrCP is a negative regulator of Wnt/beta-catenin signaling pathway and dorsal axis formation in Xenopus embryos. Mech. Dev. 77, 75–80.

Received: March 31, 2008 Revised: August 21, 2008 Accepted: October 8, 2008 Published: December 4, 2008

Millward, T.A., Zolnierowicz, S., and Hemmings, B.A. (1999). Regulation of protein kinase cascades by protein phosphatase 2A. Trends Biochem. Sci. 24, 186–191.

REFERENCES

Morin, P.J., Sparks, A.B., Korinek, V., Barker, N., Clevers, H., Vogelstein, B., and Kinzler, K.W. (1997). Activation of beta-catenin-Tcf signaling in colon cancer by mutations in beta-catenin or APC. Science 275, 1787–1790.

Bienz, M., and Clevers, H. (2000). Linking colorectal cancer to Wnt signaling. Cell 103, 311–320.

Mumby, M.C., Russell, K.L., Garrard, L.J., and Green, D.D. (1987). Cardiac contractile protein phosphatasese. J. Biol. Chem. 262, 6257–6265.

Behrens, J., Jerchow, B.A., Wurtele, M., Grimm, J., Asbrand, C., Wirtz, R., Kuhl, M., Wedlich, D., and Birchmeier, W. (1998). Functional interaction of an axin homolog, conductin, with beta-catenin, APC, and GSK3beta. Science 280, 596–599.

Munemitsu, S., Albert, I., Souza, B., Rubinfeld, B., and Polakis, P. (1995). Regulation of intracellular b-catenin levels by the adenomatous polyposis coli (APC) tumor-suppressor protein. Proc. Natl. Acad. Sci. U.S.A. 92, 3046–3050.

Bhattacharya, G., and Boman, B.M. (1995). Phosphorylation of the adenomatous polyposis coli protein and its possible regulatory effects in cells. Biochem. Biophys. Res. Commun. 208, 103–110.

Nakamura, T., Hamada, F., Ishidate, T., Anai, K., Kawahara, K., Toyoshima, K., and Akiyama, T. (1998). Axin, an inhibitor of the Wnt signalling pathway, interacts with b-catenin, GSK-3b and APC and reduces the b-catenin level. Genes Cells 3, 395–404.

660 Molecular Cell 32, 652–661, December 5, 2008 ª2008 Elsevier Inc.

Molecular Cell Regulation of b-Catenin by APC

Neufeld, K.L., Zhang, F., Cullen, B.R., and White, R.L. (2000). APC-mediated downregulation of beta-catenin activity involves nuclear sequestration and nuclear export. EMBO Rep. 1, 519–523. Polakis, P. (1997). The adenomatous polyposis coli (APC) tumor suppressor. Biochim. Biophys. Acta 1332, F127–F147.

Su, L.K., Vogelstein, B., and Kinzler, K.W. (1993). Association of the APC tumor suppressor protein with catenins. Science 262, 1734–1737. Tetsu, O., and McCormick, F. (1999). Beta-catenin regulates expression of cyclin D1 in colon carcinoma cells. Nature 398, 422–426.

Polakis, P. (2007). The many ways of Wnt in cancer. Curr. Opin. Genet. Dev. 17, 45–51.

Vogelstein, B., and Kinzler, K.W. (2002). The Genetic Basis of Human Cancer, Second Edition (New York: McGraw-Hill).

Rosin-Arbesfeld, R., Townsley, F., and Bienz, M. (2000). The APC tumour suppressor has a nuclear export function. Nature 406, 1009–1012.

Xing, Y., Clements, W.K., Kimelman, D., and Xu, W. (2003). Crystal structure of a beta-catenin/axin complex suggests a mechanism for the beta-catenin destruction complex. Genes Dev. 17, 2753–2764.

Rubinfeld, B., Souza, B., Albert, I., Muller, O., Chamberlain, S.H., Masiarz, F.R., Munemitsu, S., and Polakis, P. (1993). Association of the APC gene product with beta-catenin. Science 262, 1731–1734. Rubinfeld, B., Tice, D.A., and Polakis, P. (2001). Axin-dependent phosphorylation of the adenomatous polyposis coli protein mediated by casein kinase 1epsilon. J. Biol. Chem. 276, 39037–39045. Salic, A., Lee, E., Mayer, L., and Kirschner, M.W. (2000). Control of betacatenin stability: reconstitution of the cytoplasmic steps of the Wnt pathway in Xenopus egg extracts. Mol. Cell 5, 523–532. Shih, I.M., Yu, J., He, T.-C., Bert Vogelstein, B., and Kinzler, K.W. (2000). The b-catenin binding domain of adenomatous polyposis coli is sufficient for tumor suppression. Cancer Res. 60, 1671–1676.

Xing, Y., Clements, W.K., Le Trong, I., Hinds, T.R., Stenkamp, R., Kimelman, D., and Xu, W. (2004). Crystal structure of a beta-catenin/APC complex reveals a critical role for APC phosphorylation in APC function. Mol. Cell 15, 523–533. Yap, A.S., Brieher, W.M., Pruschy, M., and Gumbiner, B.M. (1997). Molecular and functional analysis of cadherin-based adhesions. Annu. Rev. Cell Dev. Biol. 13, 119–146. Yang, J., Zhang, W., Evans, P.M., Chen, X., He, X., and Liu, C. (2006). Adenomatous polyposis coli (APC) differentially regulates beta-catenin phosphorylation and ubiquitination in colon cancer cells. J. Biol. Chem. 281, 17751–17757.

Molecular Cell 32, 652–661, December 5, 2008 ª2008 Elsevier Inc. 661