The Third 20 Amino Acid Repeat Is the Tightest Binding Site of APC for β-Catenin

doi:10.1016/j.jmb.2006.04.064

J. Mol. Biol. (2006) 360, 133–144

The Third 20 Amino Acid Repeat Is the Tightest Binding Site of APC for β-Catenin Jing Liu 1,3 , Yi Xing 1 , Thomas R. Hinds 2 , Jie Zheng 4 and Wenqing Xu 1 ⁎ 1

Department of Biological Structure, University of Washington, Seattle, WA 98195, USA 2

Department of Pharmacology, University of Washington, Seattle, WA 98195, USA 3

Biomolecular Structure and Design Program, University of Washington, Seattle, WA 98195, USA 4

Department of Structural Biology, St. Jude Children's Research Hospital, Memphis, TN 38105, USA

Adenomatous polyposis coli (APC) plays a critical role in the Wnt signaling pathway by tightly regulating β-catenin turnover and localization. The central region of APC is responsible for APC-β-catenin interactions through its seven 20 amino acid (20aa) repeats and three 15 amino acid (15aa) repeats. Using isothermal titration calorimetry, we have determined the binding affinities of β-catenin with an APC 15aa repeat fragment and each of the seven 20aa repeats in both phosphorylated and unphosphorylated states. Despite sequence homology, different β-catenin binding repeats of APC have dramatically different binding affinities with β-catenin and thus may play different biological roles. The third 20aa repeat is by far the tightest binding site for β-catenin among all the repeats. The fact that most APC mutations associated with colon cancers have lost the third 20aa repeat underlines the importance of APC–β-catenin interaction in Wnt signaling and human diseases. For every 20aa repeat, phosphorylation dramatically increases its binding affinity for β-catenin, suggesting phosphorylation has a critical regulatory role in APC function. In addition, our CD and NMR studies demonstrate that the central region of APC is unstructured in the absence of β-catenin and Axin, and suggest that β-catenin may interact with each of the APC 15aa and 20aa repeats independently. Published by Elsevier Ltd.

*Corresponding author

Keywords: adenomatous polyposis coli (APC); β-catenin; isothermal calorimetry; unstructured protein; phosphorylation

Introduction The tumor suppressor adenomatous polyposis coli (APC) is considered to be a gatekeeper in colorectal tumourigenesis. 1,2 Truncational mutations of APC are found in Familial Adenomatous Polyposis (FAP) and more than 80% of sporadic colonic tumors. 3–7 In addition to its roles in cytoskeletal and cell adhesion regulation,8,9 it is well established that APC plays an essential role in the Wnt-regulated degradation of β-catenin.10–14 APC encodes a large 310 kDa protein with multiple domains. The central region of APC includes three 15 amino acid (15aa) repeats and seven 20 amino acid (20aa) repeats.15,16 Mutations that truncate APC in the 20aa repeat region, such as that in SW480 colon cancer cells, lead to the accumulation of high Abbreviations used: APC, adenomatous polyposis coli; aa, amino acids; ITC, isothermal titration calorimetry; CD, circular dichroism; NOE, nuclear Overhauser effect. E-mail address of the corresponding author: [email protected] 0022-2836/$ - see front matter. Published by Elsevier Ltd.

levels of β-catenin.17 When full-length APC or the central region of APC were introduced into SW480 cells, β-catenin levels were reduced.12 Furthermore, human APC2, an APC analogue containing five 20aa repeats and no 15aa repeats, has also been shown to interact with β-catenin and can decrease β-catenin levels and signaling activity in SW480 cells.18,19 All these results suggested a critical role for the APC 20aa repeats in β-catenin turnover. Each of the APC 15aa repeats, which is not regulated by phosphorylation, binds to the structural groove formed by β-catenin armadillo repeats 5– 10.20 In comparison, every APC 20aa repeat region contains a highly conserved 20aa sequence with potential phosphorylation sites in a consensus motif SXXSSLSXLS (Figure 1(a)). Phosphorylation of APC by CK1ε and GSK-3β enhances its ability to bind and down-regulate β-catenin.21,22 In vitro isothermal calorimetric (ITC) analysis showed that phosphorylation of the third APC 20aa repeat increases its βcatenin binding affinity by ∼300-fold.23,24 Recently the crystal structures of phosphorylated APC 20aa repeat 3 in complex with β-catenin revealed that one single APC 20aa repeat together with its flanking

134

ITC Study of β-Catenin/APC Interaction

Figure 1. Structural basis of the interaction between β-catenin and APC 20aa repeats. (a) Sequence alignment of Tcf, E-cadherin and APC 20 amino acid repeats. The conserved 20aa repeat regions of APC are framed by a black rectangle. Conserved Ser residues are shown in green. The two acidic residues that may interact with the charged buttons of β-catenin are highlighted in red.25 Residues phosphorylated in the β-catenin/phospho-APC 20aa repeat and the β-catenin/phosphoE-cadherin complex structures23,24,26 are boxed in orange. The secondary structure was labeled as observed in the APC-R3 crystal structures. (b) Crystal structure of the β-catenin in complex with a phosphorylated APC-R3. Two views of the overall β-catenin/pAPC-R3 complex23,24 are related by a 180 degree rotation. The armadillo repeat region of β-catenin and p-APCR3 are shown in green and red, respectively. Asp1486, the APC residue that forms a critical salt-bridge with the first charged button of β-catenin, and five residues that are phosphorylated in vitro, are shown in red sticks. Residue F1515 was also labeled to indicate the position where we made a nonsense mutation to test the contribution of the C-terminal flanking region.

ITC Study of β-Catenin/APC Interaction

residues is packed into the entire structural groove of β-catenin (Figure 1(b)).23,24 The conserved central 20 amino acids contain multiple phosphorylation sites and bind to the groove of β-catenin armadillo repeats 1–5. It was surprising that flanking amino acids N-terminal to these conserved 20 amino acids bind along most of the structural groove of β-catenin armadillo repeats 5–12, in a conformation almost identical to that of Tcf and E-cadherin.25,26 In this Nterminal flanking region, APC residues Asp1486 and Glu1494 form salt-bridges with Lys435 and Lys312 on β-catenin, two residues that are also essential for Tcf and cadherin interactions (referred to as ‘‘charged buttons’’).23,25,26 C-terminal to the conserved APC third 20 amino acids (APC-R3), there is a short region that provides additional β-catenin interactions. Because the 15aa repeat binding site on β-catenin overlaps with that of the 20aa repeat,20,23,24 it is likely that one β-catenin molecule can only interact with one 15aa or 20aa repeat. Since all APC 20aa repeats contain a conserved region, it is generally assumed that each of the seven 20aa repeats binds one β-catenin molecule in the same manner as observed for the third APC 20aa repeat. However, both the N-terminal and C-terminal flanking regions are poorly conserved among APC 20aa repeats. For example, the Tcf-like charged-button residues are not conserved among some of the 20aa repeats (Figure 1(a)). Therefore, it is possible that different APC 20aa repeats behave differently in βcatenin interaction. It remains to be tested whether the flanking regions of conserved APC 20aa repeat region make major contributions to APC–β-catenin interactions, and whether the phosphorylation of each of APC 20aa repeat regulates its interaction with βcatenin. In addition, it is unclear if these APC 20aa repeats are spatially organized around a folded struc-

135 ture, or exist in a largely unstructured environment in the absence of APC binding partners. To provide a more complete picture of the APC–β-catenin interaction, we have analyzed the structure of APC using circular dichroism (CD) and NMR, and measured the binding thermodynamic parameters of β-catenin with one 15aa repeat fragment and seven individual 20aa repeats in both phosphorylated and unphosphorylated states, using isothermal titration calorimetry (ITC). In addition, based on the crystal structures, we analyzed the critical APC residues in the third 20aa repeat involved in its interaction with β-catenin.

Results Preparation of β-catenin and APC fragments To study the interaction between β-catenin and different 15aa and 20aa repeats of APC, we overexpressed full-length human β-catenin and various APC fragments in Escherichia coli. For convenience, here, we term the three APC 15aa repeats as APC-rA to APC-rC and seven APC 20aa repeats as APC-R1 to APC-R7, respectively. Accordingly, APC-rBC and APC-R2-R4 represent the APC fragments containing the second and third 15aa repeats and the second to fourth 20aa repeats (including the first SAMP repeat), respectively. After affinity and ion-exchange purification, we were able to obtain every APC fragment and full-length human β-catenin protein pure enough for further analysis (Figure 2(a)). Some of the APC fragments have a very low expression yield. For example, the final yield of purified APCR2 was 0.03 mg/l. To understand how phosphorylation affects the interaction between β-catenin and

Figure 2. Purification and in vitro phosphorylation of APC fragments. (a) Analysis of the purified APC protein fragments by SDS-PAGE. (b) Phosphorylation reaction of APC-R3 by CKI and GSK-3β. SDS-PAGE was used to monitor the process of the phosphorylation reaction of APC-R3. Lane 1 shows the molecular weight marker labeled in kilodaltons. APC-R3 (unphosphorylated, lane 2) displays multiple band shifts after 1 h reaction (lane 3), and further band shift after another 2 h (lane 4) and 4 h reaction (lane 5). Phosphorylation of other APC 20aa repeats has similar band-shift patterns.

136 different APC 20aa repeats, we treated each APC 20aa repeat (APC-R1 to APC-R7) and APC-R2-R4 with CKI and GSK-3β, two protein kinases that coexist with APC in the β-catenin destruction complex. The APC phosphorylation process was monitored by 4%–20% Tris-HCl SDS-PAGE gel electrophoresis (BioRad). The APC band on the SDS gel was shifted progressively upon kinase treatment, indicating multiple phosphorylation sites within individual APC 20aa repeats (Figure 2 (b)). All APC and β-catenin samples were further purified individually by size-exclusion chromatography for CD, NMR and ITC studies. Structural analysis of APC fragments by CD and NMR The central region of APC (residues ∼1000–∼2100) is highly hydrophilic. It contains only a total of 18.9% of hydrophobic residues, but 20.0% of serine/ threonine and 8.0% of proline. The folding propensity of this region is predicted using various programs, including FoldIndex, PONDR and DisEMBL.27–30 These programs suggested that the central region APC is largely unstructured. The result from the FoldIndex is shown in Figure 3(a). One APC region with low probability to form folded structure but still worth testing is located between the second and fourth 20aa repeats. To determine if the central region of APC has a defined secondary structure, the CD spectra of various purified APC fragments, including APC-R2-R3 and APC-R2-R4, were measured. The CD spectra measured for APCR2-R4 at 4 °C and 25 °C are essentially identical and feature a single minimum in molar residue ellipticity at ∼200 nm (Figure 3(b)). These spectra, as well as the CD spectra for other APC fragments (data not shown), indicate that all these APC fragments are mostly unstructured. To exclude the possibility that the APC fragment has a defined structure without regular secondary structure, we measured both 1D and 2D NMR spectra of APC-R2-R4, the longest APC fragment that we were able to obtain in large quantity and that has good solubility. 1D NMR shows no obvious chemical shift signal from 5.5 ppm to 8.5 ppm. The poor dispersion of the resonances indicates that the fragment does not have a folded structure (Figure 3 (c)). This lack of folded conformation is also demonstrated in the 2D Nuclear Overhauser Effect (NOE) spectra; there is no inter-residue NOE observed in the 2D-NOESY spectrum, which is characteristic of an unfolded structure (data not shown).31 Calorimetric analysis of APC–β-catenin interaction Various APC fragments were added to full-length β-catenin solution to measure their binding affinities and corresponding thermodynamic parameters. Protein concentrations were carefully measured by the UV absorbance of guanadinium chloride denatured proteins. The results demonstrated that each of

ITC Study of β-Catenin/APC Interaction

the APC 20aa repeats interacts with β-catenin in a 1:1 molar ratio. Despite the sequence homology, each of the seven 20aa repeats binds to β-catenin with dramatically different affinities. When unphosphorylated, APC-R3 can bind β-catenin with a Kd of ∼0.2 μM. The binding of β-catenin with every other unphosphorylated 20aa repeat is at least 60 times weaker than APC-R3, and unlikely to interact with β-catenin at physiological conditions (see Discussion). Due to the solubility limitation of βcatenin, we were only able to measure the binding affinities with a β-catenin concentration of 20 μM or lower. Under these conditions, the calculated cvalues of binding isotherms (c = Kb[P]N; [P] = βcatenin concentration; N = binding stoichiometry) of APC-R1, APC-R6 and APC-R7 were far below 1. This indicates that the binding affinities of β-catenin with APC-R1,-R6 or -R7 are much weaker than 20 μM, the β-catenin concentration used in the experiment (Table 1). The binding dissociation constant between β-catenin and APC-R5 is about 80 μM. Because of the very low protein expression yield, the binding with unphosphorylated APC-R2 was not measured. However, it can be predicted that unphosphorylated APC-R2 does not interact with β-catenin (see below). The binding affinities of β-catenin with the second and third APC 15aa repeats are between 0.1–0.8 μM, comparable to the unphosphorylated APC-R3. Our ITC experiments also reveal that phosphorylation dramatically increases the binding affinity between β-catenin and every tested APC fragment. Phospho-APC-R3 binds β-catenin 140 times tighter than unphosphorylated APC-R3 (comparable with the ∼300-fold measured with slightly different APC fragments,23,24 and phosphorylation enhances the interaction of APC-R5 with β-catenin about 1500 times (Figure 4). Similarly, other APC 20aa fragments all exhibit dramatically higher binding affinity with β-catenin upon phosphorylation (Table 1). However, APC-R3 is obviously the repeat having the highest affinity with β-catenin in both phosphorylated and unphosphorylated states. Under the same experimental condition, the Kd of β-catenin/phospho-APC-R3 is 1.5 nM, and all other repeats when phosphorylated bind β-catenin 20 to 300 times weaker than phospho-APC-R3 (Table 1). For example, phosphorylated APC-R1, APC-R4, APC-R5, APC-R6 and APC-R7 have dissociation constants of 82 nM, 27 nM, 52 nM, 409 nM and 190 nM, respectively. In our ITC study, there was no detectable signal when 80 μM phospho-APC-R2 was injected into 8 μM β-catenin solution (Figure 4(b)). This indicates that the APC-R2 fragment essentially does not interact with β-catenin. The APC-rBC fragment interacts with two βcatenin molecules, indicating each 15aa repeat binds to one β-catenin molecule. In addition, APC-R2-R4 also interacts with two molecules of β-catenin, indicating that both APC-R3 and APC-R4, but not APC-R2, interact with one molecule of β-catenin simutaneously (Table 1). Thus our ITC study confirms the binding stoichiometry of one β-catenin binding repeat of APC to one β-catenin molecule.

ITC Study of β-Catenin/APC Interaction

137

Figure 3. The central region of APC is unstructured. (a) Disordered segments of APC central region (1000–2100) predicted by FoldIndex (http://www.bip.weizmann.ac.il/fldbin/findex). Similar results were obtained using DisEMBL and Pondr (data not shown). The positions of 15aa and 20aa repeats (including their flanking regions as used in this study) are aligned with corresponding residue numbers of the central region of APC. The mutation cluster region (MCR) of APC is also shown in the corresponding location. (b) The far-ultraviolet (UV) CD spectra of APC-R2-R4, measured at 4 °C and 25 °C. (c) 1H-1D spectrum of APC-R2-R4. The water signal was suppressed by using a pre-saturation scheme.

ITC Study of β-Catenin/APC Interaction

138

Table 1. Summary of isothermal calorimetric analysis of interactions between β-catenin and APC fragments

p-APC-R1 p-APC-R2 p-APC-R3 p-APC-R4 p-APC-R5 p-APC-R6 p-APC-R7 APC-R1 APC-R3 APC-R4 APC-R5 APC-R6 APC-R7 p-APC-R2-R4 site 1 p-APC-R2-R4 site 2 APC-rBC site 1 APC-rBC site 2

ΔH (Kcal/mol)

TΔS (Kcal/mol)

ΔG (Kcal/mol)

Kd (nM)

−24.5 ± 0.2 n.s. −22.5 ± 0.1 −7.2 ± 0.1 −21.9 ± 0.2 −8.7 ± 0.8 −9.8 ± 0.1 n.s. −16.8 ± 0.1 −6.7 ± 0.6 −5.5 ± 0.8 n.s. n.s. −13.0 ± 0.1 −13.3 ± 0.4 −27.7 ± 0.1 −36.4 ± 0.2

−15.0 n.a. −10.6 2.9 −12.1 −0.10 −0.75 n.a. −7.8 −0.07 −0.15 n.a. n.a. −1.1 −3.8 −19.5 −27.1

−9.5 n.a. −11.9 −10.1 −9.8 −8.6 −9.1 n.a. −9.0 −6.6 −5.4 n.a. n.a. −11.9 −9.5 −8.2 −9.3

82( ± 7.7) ≫8000 (est.) a 1.5( ± 0.6) 27( ± 14) 52( ± 4.0) 409( ± 99) 190( ± 15) ≫21,000 (est.) a 210( ± 7.0) 12,000( ± 2,700) 80,000( ± 3,200) ≫14,000 (est.) a ≫18,000 (est.) a 1.4 ± 0.7 86 ± 13 800( ± 180) 110( ± 8.1)

All the experiments were performed at T = 293 K. n.s., no detectable signal. n.a., not available. p-, phosphorylated form of APC fragments. a The Kd is estimated to far exceed the β-catenin concentration used in the experiment.

Mutational analysis on the APC-R3 and β-catenin interface The conserved 20 amino acids in the APC-R3 fragment contain three conserved residues S1504, S1507 and S1510 that can be phosphorylated by CKI in vitro.23,24 In the crystal structure, these three phosphoserine residues of APC are sandwiched by the R335 and R292 of β-catenin (Figure 1(b)). Comparing to the affinity of phosphorylated APCR3 (1.5 nM), that of the phosphorylated S1507A mutant is reduced by tenfold (15 nM, Table 2). In comparison, unphosphorylated S1507A mutant has essentially the same binding affinity with the wildtype. In the crystal structure, S1505 of APC is also phosphorylated and forms a salt-bridge with a Cterminal APC residue (R1523). However, phosphorylated S1505 does not form a hydrogen bond with βcatenin.23,24 Our ITC experiment showed that S1505A did not significantly affect the binding affinity of APC-R3 to β-catenin, either in the phosphorylated state or unphosphorylated state. We did not test how the phosphorylation of APC S1504 and S1510 may affect the β-catenin–APC-R3 interaction, since these two residues are not conserved in other 20aa repeats. In the β-catenin/APC-R3 crystal structure, the Nterminal flanking region interacts with β-catenin armadillo repeats 5–12, in a conformation similar to that of Tcf and cadherin. To understand the relative contribution of the N-terminal flanking regions of APC-R3 to β-catenin interaction, we mutated APC D1486 that forms a salt-bridge with the first charged button (K435) on the β-catenin surface. For both Tcf and E-cadherin, corresponding salt-bridges make the most critical contribution to their interaction with β-catenin.25,26 The ITC experiment shows that the phosphorylated and unphosphorylated APC-R3 (D1486A) bind to β-catenin with affinities four and 30-fold weaker than the wild-type, respectively. Thus this charged button makes a major contribution to stabilize the interaction between β-catenin

and APC-R3 in the unphosphorylated but not the phosphorylated state. APC residues 1514–1529 are C-terminal to the conserved 20 amino acids of APC-R3. In the crystal structure, this C-terminal tail fits into the groove formed by armadillo repeats 2–3 and makes some direct interactions with β-catenin (Figure 1(b)).23,24 To determine the importance of this C-terminal tail, we produced an APC-R3 fragment truncated at position F1515. Our ITC measurement demonstrated that APC-R3 truncated at F1515 (F1515X) had two and fourfold decrease in the phosphorylated and unphosphorylated states, respectively. Thus the tail region of APC-R3 makes only minor contributions to the binding of β-catenin.

Discussion Structure of the central region of APC APC is a tumor suppressor linked to FAP (familial adenomatous polyposis coli) and to the initiation of sporadic human colorectal cancer.1,2,32 The majority of the colorectal cancers have truncation mutations in the APC gene. The APC protein contains three 15aa repeats and seven 20aa repeats that are involved in its interaction with β-catenin.15,16 We measured the CD spectra of various purified individual APC repeats which demonstrated that each of these APC 15aa and 20aa repeats does not contain significant α-helix or β-sheet content (data not shown). In addition, our NMR and CD studies of the 42 kDa APC-R2-R4 fragment, which contains the second to the fourth 20aa repeats and the first Axininteracting SAMP repeat, demonstrated that this fragment does not contain a folded structure. These results are consistent with the observation that the entire central region of APC has a very low content of hydrophobic residues, suggesting a lack of a

ITC Study of β-Catenin/APC Interaction

139

Figure 4. Calorimetric analysis of the binding of different APC protein fragments to the full-length human β-catenin. The raw data obtained upon each injection were measured in μcal/s and plotted in the inset windows over the timecourse of the experiment. The binding isotherms were graphed and normalized to the concentration of the injected protein. (a) 5.6 μM β-catenin titrated with 51.6 μM phosphorylated APC-R1. (b) 8.9 μM β-catenin titrated with 88.2 μM phosphorylated APC-R2. (c) 8.4 μM β-catenin titrated with 85.0 μM phosphorylated APC-R3. (d) 1.8 μM β-catenin titrated with 17.6 μM phosphorylated APC-R4. (e) 4.3 μM β-catenin titrated with 35.4 μM phosphorylated APC-R5. (f) 1.9 μM βcatenin titrated with 19.4 μM phosphorylated APC-R6. (g) 6.8 μM β-catenin titrated with 77.7 μM phosphorylated APCR7. (h) 32.7 μM β-catenin titrated with 166 μM 15aa repeat APC-rBC.

ITC Study of β-Catenin/APC Interaction

140

Table 2. Summary of ITC of full-length human β-catenin interaction with mutants of APC 20aa repeat 3

p-APC-R3 p-APC-R3 D1486A p-APC-R3 S1505A p-APC-R3 S1507A p-APC-R3 F1515X APC-R3 APC-R3 D1486A APC-R3 S1505A APC-R3 S1507A APC-R3 F1515X

ΔH (Kcal/mol)

TΔS (Kcal/mol)

ΔG (Kcal/mol)

Kd (nM)

−22.5 ± 0.1 −17.0 ± 0.2 −23.4 ± 0.2 −24.3 ± 0.3 −15.9 ± 0.2 −16.8 ± 0.1 −8.7 ± 0.3 −15.5 ± 0.2 −25.0 ± 0.1 −12.6 ± 0.1

−10.6 −5.9 −11.3 −13.8 −4.4 −7.8 −1.8 −6.4 −16.1 −4.5

−11.9 −11.1 −12.1 −10.5 −11.5 −9.0 −6.9 −9.1 −8.9 −8.1

1.5( ± 0.6) 5.6( ± 1.3) 1.1( ± 0.1) 15( ± 4.0) 3.2( ± 1.5) 210( ± 7.0) 6800( ± 500) 189( ± 23) 240( ± 12) 858( ± 54)

All the experiments were performed at T = 293 K. p-, phosphorylated.

hydrophobic structural core (Figure 3(a)). Although it remains to be tested in the context of full-length APC protein, it is likely that the entire central region of APC is structurally flexible in the absence of its binding partners, such as β-catenin and/or Axin. Since the N-terminal region of APC contains a wellfolded oligomerization domain, a coiled-coil domain and an armadillo repeat domain,33,34 we cannot exclude the possibility that the N and/or C-terminal APC regions provide a structural scaffold for organizing part of the otherwise unstructured central region of APC. However, because of the large size of the central region of APC (>120 kDa), the induced-folding of the entire APC central region is very unlikely. Distinct β-catenin binding properties among different APC 20aa repeats Despite the presence of a highly conserved 20aa segment, our ITC measurement revealed that different APC 20aa repeats bind to β-catenin with dramatically different binding affinities. In the unphosphorylated state, while the binding of APCR1, APC-R2, APC-R6 and APC-R7 was undetectable, the β-catenin binding affinities of APC-R3, APC-R4 and APC-R5 are 0.2 μM, 12 μM and 81 μM, respectively. Such striking difference in binding affinities may be largely due to the N-terminal flanking region of the conserved 20 amino acid segment as they have distinct sequences. In addition, the crystal structure of β-catenin in complex with unphosphorylated APC-R3 showed that the Nterminal flanking region of APC-R3 but not the conserved 20aa repeat region interacts with βcatenin.23,24 The N-terminal flanking sequences of APC-R3, R4 and R5 fit the typical Tcf-like β-catenin binding motif DxθθxΦxxE, with x, θ and Φ indicating any amino acid, hydrophobic, and aromatic residues, respectively.23–25,35–39 The first Asp residue of this sequence, which is present in APC-R3, R4 and R5, but not other APC 20aa repeats (Figure 1(a)), forms a salt-bridge with β-catenin Lys435 (the first charged button) in the β-catenin/ APC-R3 crystal structures. Corresponding residues in Tcf and E-cadherin form identical salt-bridges with β-catenin and are the most important contributor for binding affinity.25,26 Mutation of the cor-

responding Asp residue in Tcf4 to Ala reduced the binding affinity by two orders of magnitude.37,38,40 While APC-R1 and APC-R6 contains a less favorable Glu in the corresponding position, APC-R2 and APC-R7 have Pro and Tyr in the corresponding position. Therefore, our results indicate that the Nterminal flanking region of any of the unphosphorylated APC 20aa repeats determines its binding affinities with β-catenin. Regulation of APC/β-catenin interaction by APC phosphorylation Consistent with previous studies with the third 20aa repeats,23,24 the phosphorylation of any of the 20aa repeat fragments increased the binding affinities by at least 140-fold (Table 1). This suggests that the function of all 20aa repeats, and likely the entire central region of APC, is tightly regulated by phosphorylation. It should be noted that the physiological phosphorylation sites on APC has not been characterized, mainly because of the technical difficulty due to the large number of Ser/ Thr residues in the APC 20aa repeat region. Previous in vitro studies on APC-R3 have revealed the Ser residues in the conserved S*S*LS*ALS* motif as phosphorylation sites by CK1 and GSK3β.21,23,24,41 However, phosphorylation can also occur beyond this conserved motif (e.g. T1487 in APC-R3; Figure 1(a)). Since the core 20aa repeat region is highly conserved, corresponding sites in other APC 20aa repeats may be also phosphorylated by CK1 and GSK-3β (Figure 1(a)). Although the in vitro phosphorylation sites in this work are not necessarily exactly the same as the in vivo sites, some of these sites should be meaningful, since CKI and GSK-3β are two kinases in the β-catenin destruction complex and have been shown to play a role in APC phosphorylation. Even if not all in vitro sites are physiologically relevant, our overall conclusion that phosphorylation drastically enhances the β-catenin–APC interaction should hold. In addition, consistent with the Kd values measured in the unphosphorylated state, APC-R3 is also the tight β-catenin binding site in the phosphorylated state. Our result is consistent with previous reports that unphosphorylated APC-R6 and R7 does not interact with β-catenin in GST-pulldown and competition

ITC Study of β-Catenin/APC Interaction

assays,24 and that in the unphosphorylated state, APC-R3 binds tighter than APC-R4 in a previous ITC assay.42 Furthermore, our result that APC-R2, even in the phosphorylated state, does not interact with β-catenin at all is consistent with the fact that the N-terminal region of APC-R2 bears no resemblance to the DxθθxΦxxE sequence. In the crystal structure of a p-APC-R2-R3/β-catenin complex, only the APC-R3 region, but not APC-R2, was found in complex with β-catenin.23 This also agrees well with the previous study that an APC fragment containing APC-R2 (residues 1342–1476) does not bind β-catenin at all.41 Potential functional differences of different APC 20aa repeats The majority of APC mutations in colon cancer occur in the mutation cluster region (MCR) between residue 1250 and residue 1450,1,2,32,43 i.e. between APC-R1 and the beginning of APC-R3. Thus most APC mutations associated with colon cancers have lost not only all SAMP repeats, the Axin binding sites, but also APC-R3, the major β-catenin binding site. When measured in Xenopus egg extracts in the absence of a Wnt signal, β-catenin concentration is around 25 nM. Upon Wnt stimulation, cellular βcatenin concentration increases to approximately 150 nM.44 Different cell types are likely to have different β-catenin concentrations in different physiological conditions, and the local concentration of APC and β-catenin could be drastically different. Yet it is likely that the binding affinities between βcatenin and most unphosphorylated APC 20aa repeats (except APC-R3) are way too low to be physiologically relevant, and the phosphorylation of 20aa repeat is critical for the function of APC 20aa repeats. It is thus plausible that different APC repeats may be involved in the interaction with βcatenin in different scenarios. In the HCT116 and SW480 cells, it has been shown that the molar ratio of β-catenin to APC (1034–2843), an APC fragment containing all β-catenin binding repeats except APC-rA, was approximately 1.7:1 at the highest βcatenin concentration used.45 This is consistent with the idea that only a small fraction of APC β-catenin binding repeats are involved in β-catenin interaction in the cell. APC may play multiple roles in the canonical Wnt pathway. APC plays a role in the nuclear export of β-catenin.46–50 In this process, APC may need to compete for β-catenin interaction with Tcf , a βcatenin partner with low nM affinity. The only βcatenin binding site of APC that has comparable binding affinity with Tcf is the phosphorylated APCR3, suggesting that APC-R3 is necessary for βcatenin nuclear export. APC may bind to β-catenin and sequester β-catenin from binding to other partners.23,24,49 In this process, some weaker βcatenin binding sites, such as APC 15aa repeats and phosphorylated APC-R4, may also play a role. Another important role of APC in Wnt signaling is

141 in the β-catenin destruction complex. In this complex, Axin acts as a scaffold protein to recruit other components to close proximity. The local concentration of APC and β-catenin in this complex might be high enough to enable other lower affinity 20aa repeats to interact with β-catenin and thus contribute to the β-catenin turnover.51 Dissection of the β-catenin–APC-R3 interaction The third 20aa repeat of APC is the predominant binding site for β-catenin. In agreement with a previous study,42 the phosphorylation site mutation S1507A in the conserved 20aa motif decreases phosphorylated state affinity ten times and has essentially no change in the unphosphorylated state. A previous study showed that the point mutation S1505D dramatically increased the binding affinity of unphosphorylated APC-R3 to β-catenin. However, in our study, no significant affinity change was observed for the either phosphorylated or unphosphorylated APC-R3 mutant S1505A. This discrepancy may be explained by the structural flexibility of the conserved 20aa repeat region in the unphosphorylated state.20 For the S1505A mutation, it does not affect the binding when APC-R3 is unphosphorylated, since the whole conserved S*S*LS*ALS* motif is unstructured. When APC-R3 is phosphorylated, S1505A mutation does not affect the binding affinity since the phosphorylation of neighboring serine residues (S1504, S1507 and S1510) provides the major binding force by interacting with positively charged pockets. In comparison, the single mutation S1505D may allow the negatively charged S1505D side-chain to selectively bind to positively charged pockets of β-catenin that S1505 does not bind in wild-type unphosphorylated APC-R3. Among APC-R3 mutants, the N-terminal charged button mutation decreased the affinity of the phosphorylated state APC-R3 to β-catenin by fourfold, and unphospho-state by more than 30fold. In comparison, the truncation of the C-terminal flanking region at F1515 only has a minor effect on the β-catenin/APC-R3 binding affinity (Table 2). These ITC data and sequence alignment suggest that the central SXXSSLSXLS region contributes the most in the phosphorylated form to the binding of βcatenin, while in the unphosphorylated state the Nterminal flanking region makes the most critical contributions to the binding affinity. This is in agreement with crystal structures23,24 and the result from a recent quantitative fluorescent study that, in its unphosphorylated state, the APC-R3 N-terminal flanking fragment (1485–1499) could interact with β-catenin, while the conserved region of APC-R3 fragment (1494–1513) could not.38 In summary, our experimental results and the APC sequence analysis indicate that the entire central region of APC is structurally flexible in the absence of its binding partners. Different APC 20aa repeats have dramatically different binding affinities with βcatenin and may have different functions in Wnt signaling. The observation that APC-R3, the

142 predominant β-catenin binding repeat, is lost in most APC mutations found in colon cancers, underlines the importance of APC–β-catenin interaction in Wnt signaling and human diseases. Together, our results provide a more complete picture of β-catenin–APC interactions.

Materials and Methods Construction of plasmids and mutagenesis Using human APC as a template, the following seven different APC cDNA fragments were subcloned into the His-tag fusion vector pPROEX-HTb: APC 20aa repeat 1 (APC-R1, residues 1225–1314), APC 20aa repeat 2 (APCR2, residues 1342–1410), APC 20aa repeat 3 (APC-R3, residues 1459–1532), APC 20aa repeat 4 (APC-R4, residues 1614–1679), APC 20aa repeat 5 (APC-R5, residues 1811–1899), APC 20aa repeat 6 (APC-R6, residues 1931–1991), APC 20aa repeat 7 (APC-R7, residues 1988– 2050). Recombinant His-tagged form of full-length human β-catenin, the APC fragment containing the second and third 15aa repeats (APC-rBC, residues 1133–1189) and the APC fragment containing the second and third 20aa repeat (APC-R2-R3, residues 1362–1540) were produced as described.51 The APC fragment containing the second to fourth 20aa repeats and the first SAMP repeat (APC-R2R4, residues 1362–1745) was subcloned into glutathioneS-transferase (GST) tag vector pGEX-4T1. The mutated APC repeat 3 constructs, D1486A, S1505A, S1507A and F1515X (X: stop codon) were produced through sitedirected mutagenesis (Stratagene). All APC constructs were confirmed by DNA sequencing. Protein expression and purification All APC 20aa repeats were over-expressed in BL21 (DE3) Star cells (Novagen). Cell pellets were washed by PBS buffer and sonicated in a solution containing 20 mM Tris buffer (pH 8.0), 200 mM NaCl, 2 mM β-mercaptoethanol. The sonication supernatant was applied to a Ni2+-NTA affinity column (QIAGEN). After extensive wash with 20 mM Tris buffer (pH 8.0), 200 mM NaCl, 2 mM β-mercaptoethanol, the protein was eluted with a buffer containing 20 mM Tris (pH 8.0), 100 mM NaCl and 200 mM imidazole. All proteins were further purified by a Q-Sepharose HP column (Amersham Pharmacia). APC-R2 was sequentially purified by both Q- and SP-Sepharose HP columns (Amersham Pharmacia). The mutants of APC repeat 3 were purified in the same way as the wild-type (WT) protein. After the above steps of purification, all the APC repeats appeared as a single band in SDS-PAGE gels and were stable in solution at 4 °C for several days. They were stored in 20 mM Tris (pH 8.0), 150 mM NaCl, 3 mM dithiothreitol (DTT). The protein expression yields of APC-R2 and APC-R4 were improved by using 2×YT medium instead of LB medium, and extending the protein expression time after isopropyl β-D-thiogalactoside (IPTG) induction from 3 h to 5 h. Full-length human β-catenin (residues 1–781) used for the isothermal calorimetric binding experiments was purified as described.51 APC-rBC and APC-R2-R4 were purified by glutathione affinity chromatography followed by TEV protease cleavage of the GST-tag in TEV cleavage buffer containing 20 mM Tris (pH 8.0), 150 mM NaCl, 0.5 mM EDTA, 1 mM DTT for 16 h at room temperature.

ITC Study of β-Catenin/APC Interaction The cleaved protein was further purified by Q-Sepharose HP column (Amersham Pharmacia) and Superdex-200 size-exclusion column (Amersham Pharmacia) and finally changed to 1× PBS buffer for NMR study. In vitro phosphorylation Phosphorylated APC samples were prepared using purified APC fragments extensively phosphorylated by CK1-δ and GSK-3β (New England Biolabs). The reaction buffer contained 50 mM Tris (pH 7.5), 10 mM MgCl2, 5 mM ATP, 5 mM DTT and all the protein samples were incubated at 30 °C. Final protein and enzyme concentration were around 0.1 μM and 100 unit/μl, respectively. The reaction was monitored by the APC band-shift in SDS-PAGE gel. When the band-shift was completed, EDTA at final concentration of 20 mM was added to quench the phosphorylation reaction. While the phosphorylation of most APC repeats could reach saturation around 4–6 h (Figure 2(b)), that of APC-R2 and APC-R4 needed 16–36 h. Protein quantification Immediately after size-exclusion chromatography, solid guanidinium chloride was added into freshly purified protein samples in PBS buffer to a final concentration of 6 M. The molar extinction coefficients of the protein fragments in the denatured condition were determined from their amino acid composition as described.52 UV absorbance at 276, 278, 279, 280 and 282 nm were measured and averaged and converted to protein concentration in the unit of mg/ml. Isothermal calorimetric measurement Isothermal calorimetric titration measurements were carried out at constant 20 °C using a VP-isothermal titration calorimeter (MicroCal). The stock solution of all protein samples was exchanged into PBS buffer with 1 mM DTT using a Superdex-200 size-exclusion column (Amersham Pharmacia). Full-length human β-catenin was placed in the 1.42 ml cell of the calorimeter and was titrated with one of the APC fragments. Each sample was carefully degassed before titration. For unphosphorylated APC fragments, the titration experiment started with a 2 μl injection followed by 41 subsequent injections (7 μl) at 4 min intervals. The average unphosphorylated APC fragment concentration was around 200 μM and human βcatenin concentration used was around 20 μM. For phosphorylated APC fragments, the titration experiments were started with a 2 μl injection followed by 28 subsequent injections (10 μl) at 4 min intervals. The average phosphorylated APC concentration was around 40 μM, and human β-catenin concentration used was around 4 μM. For all APC-R3 mutants, either phosphorylated or unphosphorylated, the ITC experiments were started with a 2 μl injection followed by 28 subsequent injections (10 μl) at 4 min intervals. The data were analyzed and thermodynamic parameters were determined using Origin software package (MicroCal). Circular dichroism (CD) cpectroscopy The CD spectra of various APC fragments including APC-R2-R4 were measured at 4 °C and 25 °C with a Peltier device, using an Aviv 62A DS spectrometer flushed with

ITC Study of β-Catenin/APC Interaction N2 and a cuvette with 1 mm path length. The buffer was 10 mM phosphate (pH 7.5). Far-UV CD wavelength was scanned from 180 to 260 nm at 5 s intervals. The spectrum of the buffer was subtracted to achieve an appropriate signal-to-noise ratio. NMR spectroscopy NMR data were recorded with a Bruker Avance 800MHz spectrometer equipped with a cryo-probe operating at a proton frequency of 800.13 MHz with the carrier frequency set at the water resonance. All the experiments were performed at 25 °C. Sample concentration for NMR experiments was 0.2 mM in PBS and 10% 2H2O with 1 mM DTT-d10. For the one-dimensional proton experiment, the spectrum width was 9690 Hz with 4096 complex data points; a total of 128 scans were collected; and water suppression was achieved by the presaturation method. For the two-dimensional NOESY experiment, the spectrum width in both dimensions was 8800 Hz; the water signal was suppressed by the 3919 watergate method and the mixing time for NOE build-up was 100 ms. The spectrum was collected as 512 t1 values with 64 scans/ increment, each with 2048 complex points in t2 with a relaxation delay of 1.5 s. The data were processed and displaced by using Bruker software XWINNMR. For the one-dimensional proton spectrum, data were multiplied by an exponential which produced 10 Hz line broadening prior to Fourier transformation. For the two-dimensional NOESY spectrum, prior to Fourier transformation, the data set was apodized in both directions by a squared sine-bell and was zero filled to 1024 in the t1 dimension.

143

8. 9. 10. 11.

12.

13.

14. 15.

16. 17.

Acknowledgements We thank Dr J. Beavo for the use of calorimeter and Dr D. Kimelman for critical comments on the manuscript. This work was supported by NIH grant CA90351 to W.X. and GM61739 to J.Z., and by the American Lebanese Syrian Associated Charities to J.Z.

References 1. Powell, S. M., Zilz, N., Beazer-Barclay, Y., Bryan, T. M., Hamilton, S. R. Thibodeau, S. N. et al. (1992). APC mutations occur early during colorectal tumorigenesis. Nature, 359, 235–237. 2. Bienz, M. & Clevers, H. (2000). Linking colorectal cancer to Wnt signaling. Cell, 103, 311–320. 3. Fodde, R. (2002). The APC gene in colorectal cancer. Eur. J. Cancer, 38, 867–871. 4. Su, L. K., Kinzler, K. W., Vogelstein, B., Preisinger, A. C., Moser, A. R., Luongo, C. et al. (1992). Multiple intestinal neoplasia caused by a mutation in the murine homolog of the APC gene. Science, 256, 668–670. 5. Kinzler, K. W. & Vogelstein, B. (1996). Lessons from hereditary colorectal cancer. Cell, 87, 159–170. 6. Kinzler, K. W., Nilbert, M. C., Vogelstein, B., Bryan, T. M., Levy, D. B., Smith, K. J. et al. (1991). Identification of a gene located at chromosome 5q21 that is mutated in colorectal cancers. Science, 251, 1366–1370. 7. Joslyn, G., Carlson, M., Thliveris, A., Albertsen, H.,

18.

19.

20.

21.

22.

23.

24.

Gelbert, L., Samowitz, W. et al. (1991). Identification of deletion mutations and three new genes at the familial polyposis locus. Cell, 66, 601–613. Bienz, M. & Hamada, F. (2004). Adenomatous polyposis coli proteins and cell adhesion. Curr. Opin. Cell Biol. 16, 528–535. Nathke, I. S. (2004). The adenomatous polyposis coli protein: the Achilles heel of the gut epithelium. Annu. Rev. Cell Dev. Biol. 20, 337–366. Aberle, H., Bauer, A., Stappert, J., Kispert, A. & Kemler, R. (1997). Beta-catenin is a target for the ubiquitinproteasome pathway. EMBO J. 16, 3797–3804. Hart, M. J., de los Santos, R., Albert, I. N., Rubinfeld, B. & 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. Munemitsu, S., Albert, I., Souza, B., Rubinfeld, B. & Polakis, P. (1995). Regulation of intracellular betacatenin levels by the adenomatous polyposis coli (APC) tumor-suppressor protein. Proc. Natl. Acad. Sci. USA, 92, 3046–3050. Orford, K., Crockett, C., Jensen, J. P., Weissman, A. M. & Byers, S. W. (1997). Serine phosphorylation-regulated ubiquitination and degradation of beta-catenin. J. Biol. Chem. 272, 24735–24738. Peifer, M. & Polakis, P. (2000). Wnt signaling in oncogenesis and embryogenesis—a look outside the nucleus. Science, 287, 1606–1609. Rubinfeld, B., Souza, B., Albert, I., Muller, O., Chamberlain, S. H., Masiarz, F. R. et al. (1993). Association of the APC gene product with betacatenin. Science, 262, 1731–1734. Su, L. K., Vogelstein, B. & Kinzler, K. W. (1993). Association of the APC tumor suppressor protein with catenins. Science, 262, 1734–1737. Polakis, P. (1997). The adenomatous polyposis coli (APC) tumor suppressor. Biochim. Biophys. Acta, 1332, F127–F147. Nakagawa, H., Murata, Y., Koyama, K., Fujiyama, A., Miyoshi, Y., Monden, M. et al. (1998). Identification of a brain-specific APC homologue, APCL, and its interaction with beta-catenin. Cancer Res. 58, 5176–5181. McCartney, B. M., Dierick, H. A., Kirkpatrick, C., Moline, M. M., Baas, A., Peifer, M. & Bejsovec, A. (1999). Drosophila APC2 is a cytoskeletally-associated protein that regulates wingless signaling in the embryonic epidermis. J. Cell Biol. 146, 1303–1318. Eklof Spink, K., Fridman, S. G. & Weis, W. I. (2001). Molecular mechanisms of beta-catenin recognition by adenomatous polyposis coli revealed by the structure of an APC-beta-catenin complex. EMBO J. 20, 6203–6212. Rubinfeld, B., Albert, I., Porfiri, E., Fiol, C., Munemitsu, S. & Polakis, P. (1996). Binding of GSK3beta to the APC-beta-catenin complex and regulation of complex assembly. Science, 272, 1023–1026. Rubinfeld, B., Tice, D. A. & Polakis, P. (2001). Axindependent phosphorylation of the adenomatous polyposis coli protein mediated by casein kinase 1epsilon. J. Biol. Chem. 276, 39037–39045. Xing, Y., Clements, W. K., Le Trong, I., Hinds, T. R., Stenkamp, R., Kimelman, D. & 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. Ha, N. C., Tonozuka, T., Stamos, J. L., Choi, H. J. & Weis, W. I. (2004). Mechanism of phosphorylation-

ITC Study of β-Catenin/APC Interaction

144

25. 26.

27. 28.

29.

30.

31.

32. 33. 34.

35.

36.

37.

38.

dependent binding of APC to beta-catenin and its role in beta-catenin degradation. Mol. Cell, 15, 511–521. Graham, T. A., Weaver, C., Mao, F., Kimelman, D. & Xu, W. (2000). Crystal structure of a beta-catenin/Tcf complex. Cell, 103, 885–896. Huber, A. H. & Weis, W. I. (2001). The structure of the beta-catenin/E-cadherin complex and the molecular basis of diverse ligand recognition by beta-catenin. Cell, 105, 391–402. Uversky, V. N., Gillespie, J. R. & Fink, A. L. (2000). Why are ‘‘natively unfolded’’ proteins unstructured under physiologic conditions? Proteins, 41, 415–427. Prilusky, J., Felder, C. E., Zeev-Ben-Mordehai, T., Rydberg, E. H., Man, O., Beckmann, J. S. et al. (2005). FoldIndex: a simple tool to predict whether a given protein sequence is intrinsically unfolded. Bioinformatics, 21, 3435–3438. Li, X., Romero, P., Rani, M., Dunker, A. K. & Obradovic, Z. (1999). Predicting protein disorder for N-, C-, and internal regions. Genome Inform. Ser. Workshop Genome Inform. 10, 30–40. Linding, R., Jensen, L. J., Diella, F., Bork, P., Gibson, T. J. & Russell, R. B. (2003). Protein disorder prediction: implications for structural proteomics. Structure, 11, 1453–1459. Wong, H. C., Liu, G., Zhang, Y. M., Rock, C. O. & Zheng, J. (2002). The solution structure of acyl carrier protein from Mycobacterium tuberculosis. J. Biol. Chem. 277, 15874–15880. Polakis, P. (1995). Mutations in the APC gene and their implications for protein structure and function. Curr. Opin. Genet. Dev. 5, 66–71. Day, C. L. & Alber, T. (2000). Crystal structure of the amino-terminal coiled-coil domain of the APC tumor suppressor. J. Mol. Biol. 301, 147–156. Tickenbrock, L., Cramer, J., Vetter, I. R. & Muller, O. (2002). The coiled coil region (amino acids 129–250) of the tumor suppressor protein adenomatous polyposis coli (APC). Its structure and its interaction with chromosome maintenance region 1 (Crm-1). J. Biol. Chem. 277, 32332–32338. Fasolini, M., Wu, X., Flocco, M., Trosset, J. Y., Oppermann, U. & Knapp, S. (2003). Hot spots in Tcf4 for the interaction with beta-catenin. J. Biol. Chem. 278, 21092–21098. von Kries, J. P., Winbeck, G., Asbrand, C., SchwarzRomond, T., Sochnikova, N., Dell’Oro, A. et al. (2000). Hot spots in beta-catenin for interactions with LEF-1, conductin and APC. Nature Struct. Biol. 7, 800–807. Omer, C. A., Miller, P. J., Diehl, R. E. & Kral, A. M. (1999). Identification of Tcf4 residues involved in highaffinity beta-catenin binding. Biochem. Biophys. Res. Commun. 256, 584–590. Gail, R., Frank, R. & Wittinghofer, A. (2005). Systematic peptide array-based delineation of the differential betacatenin interaction with Tcf4, E-cadherin, and adenomatous polyposis coli. J. Biol. Chem. 280, 7107–7117.

39. Choi, H. J., Huber, A. H. & Weis, W. I. (2006). Thermodynamics of beta-catenin-ligand interactions. The roles of the N- and C-terminal tails in modulating binding affinity. J. Biol. Chem. 281, 1027–1038. 40. Knapp, S., Zamai, M., Volpi, D., Nardese, V., Avanzi, N., Breton, J. et al. (2001). Thermodynamics of the high-affinity interaction of TCF4 with beta-catenin. J. Mol. Biol. 306, 1179–1189. 41. Rubinfeld, B., Albert, I., Porfiri, E., Munemitsu, S. & Polakis, P. (1997). Loss of beta-catenin regulation by the APC tumor suppressor protein correlates with loss of structure due to common somatic mutations of the gene. Cancer Res. 57, 4624–4630. 42. Tickenbrock, L., Kossmeier, K., Rehmann, H., Herrmann, C. & Muller, O. (2003). Differences between the interaction of beta-catenin with non-phosphorylated and single-mimicked phosphorylated 20-amino acid residue repeats of the APC protein. J. Mol. Biol. 327, 359–367. 43. Lamlum, H., Ilyas, M., Rowan, A., Clark, S., Johnson, V., Bell, J. et al. (1999). The type of somatic mutation at APC in familial adenomatous polyposis is determined by the site of the germline mutation: a new facet to Knudson’s ‘two-hit’ hypothesis. Nature Med. 5, 1071–1075. 44. Lee, E., Salic, A., Kruger, R., Heinrich, R. & Kirschner, M. W. (2003). The roles of APC and Axin derived from experimental and theoretical analysis of the Wnt pathway. PLoS Biol. 1, E10. 45. Rubinfeld, B., Souza, B., Albert, I., Munemitsu, S. & Polakis, P. (1995). The APC protein and E-cadherin form similar but independent complexes with alphacatenin, beta-catenin, and plakoglobin. J. Biol. Chem. 270, 5549–5555. 46. Henderson, B. R. (2000). Nuclear-cytoplasmic shuttling of APC regulates beta-catenin subcellular localization and turnover. Nature Cell Biol. 2, 653–660. 47. Rosin-Arbesfeld, R., Townsley, F. & Bienz, M. (2000). The APC tumour suppressor has a nuclear export function. Nature, 406, 1009–1012. 48. van de Wetering, M., Barker, N., Harkes, I. C., van der Heyden, M., Dijk, N. J., Hollestelle, A. et al. (2001). Mutant E-cadherin breast cancer cells do not display constitutive Wnt signaling. Cancer Res. 61, 278–284. 49. Bienz, M. (2002). The subcellular destinations of APC proteins. Nature Rev. Mol. Cell Biol. 3, 328–338. 50. Henderson, B. R. & Fagotto, F. (2002). The ins and outs of APC and beta-catenin nuclear transport. EMBO Rep. 3, 834–839. 51. Xing, Y., Clements, W. K., Kimelman, D. & 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. 52. Gill, S. C. & von Hippel, P. H. (1989). Calculation of protein extinction coefficients from amino acid sequence data. Anal. Biochem. 182, 319–326.

Edited by M. Yaniv (Received 5 January 2006; received in revised form 13 April 2006; accepted 27 April 2006) Available online 15 May 2006