Crystal structure of a full-length beta-catenin.
Journal: 2008/June - Structure
ISSN: 0969-2126
Abstract:
beta-catenin plays essential roles in cell adhesion and Wnt signaling, while deregulation of beta-catenin is associated with multiple diseases including cancers. Here, we report the crystal structures of full-length zebrafish beta-catenin and a human beta-catenin fragment that contains both the armadillo repeat and the C-terminal domains. Our structures reveal that the N-terminal region of the C-terminal domain, a key component of the C-terminal transactivation domain, forms a long alpha helix that packs on the C-terminal end of the armadillo repeat domain, and thus forms part of the beta-catenin superhelical core. The existence of this helix redefines our view of interactions of beta-catenin with some of its critical partners, including ICAT and Chibby, which may form extensive interactions with this C-terminal domain alpha helix. Our crystallographic and NMR studies also suggest that the unstructured N-terminal and C-terminal tails interact with the ordered armadillo repeat domain in a dynamic and variable manner.
Relations:
Content
Citations
(61)
References
(50)
Chemicals
(3)
Genes
(1)
Organisms
(3)
Processes
(4)
Affiliates
(1)
Similar articles
Articles by the same authors
Discussion board
Structure 16(3): 478-487

Crystal Structure of a Full-Length β-Catenin

INTRODUCTION

β-catenin was originally identified in cell adherens junctions where it functions to bridge the cytoplasmic domain of cadherins to α-catenin and the actin cytoskeleton (Hulsken et al., 1994; McCrea et al., 1991). Besides being an essential component in cell adhesion, β-catenin is also a key signaling molecule in the canonical Wnt pathway that plays diverse and critical roles in embryonic development and in adults (Moon and Kimelman, 1998; Peifer and Polakis, 2000). Deregulation of β-catenin activity is associated with cancer and other human diseases (Moon et al., 2002; Peifer and Polakis, 2000). For example, constitutive upregulation of β-catenin-dependent transcriptional activity is stimulated in most colon cancers through loss-of-function mutations in the tumor suppressors APC and Axin, or through gain-of-function mutations in β-catenin itself (Bienz and Clevers, 2000; Kinzler and Vogelstein, 1996; Polakis, 1997). Specifically, APC mutations that lead to β-catenin accumulation were found in more than 80% of colon cancers, while β-catenin is directly mutated and thus activated in the majority of hepatoblastomas and pilomatricoma (Bienz and Clevers, 2000; Moon et al., 2004; Polakis, 2000). Recently, the Wnt/β-catenin pathway has also emerged as a critical regulator of stem cells (Reya and Clevers, 2005), and has been implicated in bone-density syndromes as well as in Alzheimer’s disease (Moon et al., 2004). Therefore, β-catenin is a rationale target for drug development for multiple diseases, raising considerable interest in its structure.

The 781 amino acid β-catenin protein contains a central structural core of 12 armadillo repeats (residues 138–664) (Huber et al., 1997). The positively charged groove that spans the entire superhelical armadillo repeat region constitutes the binding surface for the majority of more than 20 β-catenin partners, many critical for cell adhesion and Wnt signaling (Daniels and Weis, 2002; Eklof Spink et al., 2001; Graham et al., 2000, 2002; Ha et al., 2004; Huber et al., 1997; Huber and Weis, 2001; Poy et al., 2001; Xing et al., 2003, 2004; Zhurinsky et al., 2000). Sequences of β-catenin terminal domains are less conserved than the armadillo repeat domain. Both the N- and C-terminal domains are sensitive to mild protease digestion and unlikely to form stably folded structure by themselves (Huber et al., 1997). Nevertheless, β-catenin terminal domains mediate a subset of protein-protein interactions, which together with the armadillo repeat domain, enables β-catenin to function as a scaffold for multiprotein assemblies. For example, the N-terminal domain of β-catenin connects the β-catenin/E-cadherin complex to α-catenin, which is a key regulator of the actin cytoskeleton (Drees et al., 2005; Nagafuchi, 2001; Yamada et al., 2005). In addition, during degradation of cytosolic β-catenin, β-catenin is ubiquitinated when its phosphorylated N terminus is recognized by the β-TrCP ubiquitin ligase (Jiang and Struhl, 1998; Wu et al., 2003).

The majority of β-catenin partners are involved in the regulation of Wnt-responsive gene transcription in the nucleus. Among them, Tcf family members interact with β-catenin armadillo repeats 3–10 and anchor β-catenin to specific promoters of transcription. In addition to the Tcf binding region, the first armadillo repeat (R1) and armadillo repeat 11 to the C terminus (R11-C) were identified as essential regions for the transactivation of Wnt target genes (Stadeli et al., 2006; Willert and Jones, 2006). β-catenin R1 interacts with BCL9 that in turn recruits Pygopus, a critical transcriptional coactivator in Wnt signaling (Kramps et al., 2002). The β-catenin R11-C region has been shown to interact with many transcriptional coactivators, such as parafibromin, Brg1, CBP/p300, MED12, which function in different stages of transcription, as well as transcriptional inhibitors such as ICAT and Chibby (Barker et al., 2001; Hecht et al., 2000; Kim et al., 2006; Mosimann et al., 2006; Sierra et al., 2006; Tago et al., 2000; Takemaru et al., 2003; Takemaru and Moon, 2000).

Intriguingly, based on yeast two-hybrid screen and GST-pull-down assays that were visualized by Western blotting, it was suggested that the terminal domains, in particular the C-terminal domain of β-catenin, may form intramolecular interactions with the armadillo repeat domain and regulate ligand binding (Castano et al., 2002; Cox et al., 1999; Piedra et al., 2001; Solanas et al., 2004). Contrary to these earlier studies that employed less specific binding assays (e.g., yeast two-hybrid), a recent calorimetric study has shown that the N- and C-terminal domains did not affect the affinity of β-catenin for tight ligands such as E-cadherin, Lef-1, and phosphorylated APC. Whether β-catenin is regulated by intramolecular interactions is thus unresolved.

Several β-catenin protein complex structures have been solved by X-ray crystallography. However, all previous structural studies have only used the β-catenin armadillo repeat domain rather than full-length β-catenin (Daniels and Weis, 2002; Eklof Spink et al., 2001; Graham et al., 2000, 2002; Ha et al., 2004; Huber et al., 1997; Huber and Weis, 2001; Poy et al., 2001; Xing et al., 2003, 2004). To understand the regulation of cell adhesion and Wnt signaling, it is critical to unravel the structure of the full-length β-catenin. Here, we report the crystal structures of the full-length zebrafish β-catenin and a human β-catenin fragment (β-catenin-R1C) that contains residues from the first armadillo repeat to the C terminus, at 3.4 and 2.2 Å resolution, respectively. Our crystal structures reveal that the conserved N-terminal region of the C-terminal domain forms an α helix that constitutes part of the structural core of β-catenin and is likely to play an important role in Wnt-responsive transcription. We show that zebrafish β-catenin N- and C-terminal tails are unstructured and do not interact with the β-catenin armadillo repeat domain with a static conformation, and the negatively charged C-terminal tail may interact with the positively charged armadillo repeat domain in a highly dynamic and variable manner. Our work provides a structural basis for understanding the various functions of β-catenin and for pursuing drug design.

RESULTS

Crystal Structure of β-Catenin-R1C

We first crystallized a human β-catenin fragment (termed as β-catenin-R1C, residues 138–781), which contains both armadillo repeat domain (residues 138–664) of β-catenin and β-catenin C-terminal domain (residues 665–781). Crystals of this β-catenin-R1C fragment contain one molecule per asymmetric unit. The structure was first solved by molecular replacement (MR). In addition to the armadillo repeat domain that is used as the search model, we instantly observed clear electron density for the first 27 residues (residues 665–691) of the β-catenin C-terminal domain, which consist of an α helix (667–683, referred to as helix C) and a loop with a 310-helical turn (684–691). But the rest of the C-terminal domain (referred to as the C-terminal tail in this work) was invisible in the electron density map from MR. To avoid model bias intrinsic to the MR method, we also determined the structure de novo by the SIRAS method and refined the structure to 2.2 Å (Table 1). The initial experimental map has easily recognizable features with excellent contrast and continuity within the protein region (Figure 1A). Nevertheless, only residues 665–691 of β-catenin C-terminal domain, which were also identified by MR method, were visible in the electron density maps before and after density modification. Further structure refinement did not reveal any more electron density belonging to the β-catenin C-terminal tail. The crystal has been proven to contain intact β-catenin-R1C fragment by the SDS electrophoresis of the dissolved crystals (Figure 1C). These data demonstrate that the C-terminal tail is flexible in the β-catenin-R1C structure.

An external file that holds a picture, illustration, etc.
Object name is nihms98359f1.jpg
Human β-Catenin-R1C Crystal Structure

(A) Stereoview of SIRAS experimental map of β-catenin-R1C C-terminal domain helix contoured at 1 σ.

(B) β-catenin-R1C crystal structure. Each armadillo repeat of β-catenin, except repeat 7, is composed of three helices that are shown as blue (helix 1), green (helix 2), and yellow (helix 3) cylinders, whereas the C-terminal domain is colored in red. The C-terminal domain α helix is named helix C.

(C) SDS-gel analysis of β-catenin-R1C crystals. Lane 1, molecular weight marker. Lane 2, purified β-catenin armadillo repeat domain protein (residues 133–665). Lane 3, purified β-catenin-R1C protein (residues 133–781). Lane 4, dissolved β-catenin-R1C crystals.

Table 1

Structure Determination and Refinement of Human β-Catenin-R1C and Full-Length Zebrafish β-Catenin

Data Collectionβ-Catenin-R1CFull-Length β-Catenin
Data setsMecury derivativeNativeNative
Space groupI422I422P212121
Cell Dimensions
a, b, c (Å)106.483, 106.483, 259.830104.229, 104.229, 259.39171.623, 112.819, 123.091
Resolution (Å)a50.0–2.5 (2.59–2.50)50.0–2.2 (2.28–2.20)30.0–3.4 (3.52–3.40)
Rmerg (%)a8.2 (53.8)7.5 (61.1)9.1 (49.6)
I/σ(I)a24.6 (2.7)21.8 (1.9)12.7 (2.2)
Completeness (%)a99.7 (99.0)96.8 (81.2)89.1 (90.3)
Redundancy9.18.53.4
Mecury sites found2
Refinement Statistics
Resolution (Å)50.0–2.230.0–3.4
Number of reflections2978311140
Rwork/Rfree (%)21.3/24.928.1/31.9
Number of Atoms
 Protein40324188
 Water280
B Factors
 Protein40.166.3
 Water34.1
Rms Deviations
 Bond length (Å)0.0060.003
 Bond angle (°)1.10.7
In parentheses are corresponding numbers for the highest resolution shell.

The α helix of β-catenin C-terminal domain (helix C) packs against otherwise exposed hydrophobic residues from the last armadillo repeat, capping the C-terminal end of the armadillo repeat region. The helix C is connected to and runs parallel to the helix 3 of armadillo repeat 12 (Figure 1B). A series of hydrophobic side chains from the helix C, Leu674, Leu678, and Leu682, fit into a hydrophobic patch formed by the second and third helices of armadillo repeat 12, coordinating tightly with Leu644, Ala652, Ala656, and Phe660 (Figures 1A, 2A, and 2C). Side chain nitrogens of Lys671 and Arg684 at each end of the helix C attach to armadillo repeats through hydrogen bonds with backbone carbonyl oxygens of Glu664 and Ser646, respectively (Figures 2A and 2C). In comparison, the other surface of the helix C is composed of hydrophilic residues exposed to solvent (Figures 2A and 2C). It is remarkable that the helix C, including the three leucine residues that dock the helix C onto the armadillo repeat domain, is highly conserved in sequence, while the rest of the C-terminal domain has a divergent sequence (Figure 2B). These observations strongly suggest that this capping helix C is a shared structural feature in the β-catenin family.

An external file that holds a picture, illustration, etc.
Object name is nihms98359f2.jpg
Interactions between β-Catenin C-Terminal Domain and Central Armadillo Repeat Domain

(A) Electrostatic surface of the β-catenin armadillo repeat domain covered by the β-catenin C-terminal domain. Residues from the armadillo repeat domain are labeled in black. β-catenin C-terminal domain is in a stick model labeled in red. Key residues of the C-terminal domain interacting with the armadillo repeat domain are highlighted in green for hydrophobic interactions and yellow for charge-charge interactions.

(B) Sequence alignment of the β-catenin C-terminal domain with CLUSTALW. Invariant residues are labeled with stars in red, strongly similar residues are labeled with (:) in green, and weakly similar residues are labeled with (.) in blue. Sequences of the C-terminal domain visible in the β-catenin-R1C crystal structure are framed. The conserved hydrophobic residues (L674, L678, and L682) that docked the helix C into the armadillo repeat domain are marked with arrows.

(C) Ligplot presentation of interactions between β-catenin residues in the C-terminal domain (purple) and the armadillo repeat domain (red). Hydrogen bonds and charge-charge interactions are designated with green dashed lines and distance in Å. A red starburst together with a red dashed line represents hydrophobic interactions.

Crystal Structure of Full-Length β-Catenin

To provide a more complete picture of β-catenin structure for understanding the mechanisms of Wnt signaling and cell adhesion, we extensively screened the crystallization conditions of full-length β-catenin from various organisms. The full-length zebrafish β-catenin was the only β-catenin that yielded single crystals. The structure of the full-length zebrafish β-catenin has been solved and refined with numerous data sets collected from different crystals (Figure 3A), with the highest resolution being 3.4 Å (Table 1). The electron density map from molecular replacement has been carefully searched. No clear electron density was observed for the two terminal domains except for the helix C in the C-terminal domain and an additional α helix flanking the N terminus of armadillo repeat 1. A shorter N-terminal α helix has been observed in the crystal structures of β-catenin in complex with Tcf-3 or the unphosphorylated E-cadherin cytoplasmic domain complex (Graham et al., 2000; Huber and Weis, 2001). The final crystal structure obtained from full-length zebrafish β-catenin protein crystals contains residues 126–681 (corresponding to human β-catenin residues 127–682; Figure 3A). For simplicity, all β-catenin residue numbers used in this work are these of corresponding human β-catenin unless specified. The presence of full-length β-catenin in the crystal has been clearly verified by the SDS electrophoresis of the washed and dissolved single crystals (Figure 3B).

An external file that holds a picture, illustration, etc.
Object name is nihms98359f3.jpg
Full-Length Zebrafish β-Catenin Crystal Structure

(A) Crystal structure of full-length zebrafish β-catenin. The color scheme of β-catenin armadillo repeats is the same as that in Figure 1. The C-terminal helix (helix C) is colored in red. The N-terminal helix is colored in blue and green with an arrow pointing to the kink. All β-catenin residue numbers shown in this figure are these of corresponding residue numbers of human β-catenin.

(B) SDS-gel analysis of the dissolved full-length zebrafish β-catenin crystals. Lane 1, molecular weight marker. Lane 2, dissolved crystals. Lane 3, purified full-length zebrafish β-catenin.

(C) Structural superposition between human β-catenin-R1C (red) and full-length zebrafish β-catenin (cyan) around helix C.

It is remarkable that the helix C was observed in both β-cate-nin-R1C and full-length β-catenin crystal structures. The helix C is unlikely a crystal packing artifact, since crystal packing environments in these two crystal structures are very different and helix C is mostly not involved in direct crystal packing. The helix C together with the armadillo repeats 10–12 from β-catenin-R1C and full-length β-catenin structures are closely superimopsable, with the Cα rmsd of the helix C (residues 667–682 in human β-catenin-R1C) being 1.26 Å (Figure 3C).

The Helix C Plays Important Roles in β-Catenin-Ligand Interactions

β-catenin participates in a variety of protein-protein networks in both Wnt-signaling pathway and cell-adherens junctions. In published β-catenin complex crystal structures, the armadillo repeat 12 of β-catenin forms part of the binding interfaces for APC, E-cadherin, and ICAT (Daniels and Weis, 2002; Graham et al., 2002; Ha et al., 2004; Huber and Weis, 2001; Xing et al., 2004). To investigate whether these β-catenin partners interact with the helix C, we superimposed the crystal structure of β-cat-enin-R1C with those of the β-catenin/ICAT, the β-catenin/APC, and the β-catenin/E-cadherin complexes (Figures 4A and 4B).

An external file that holds a picture, illustration, etc.
Object name is nihms98359f4.jpg
The Role of Helix C in β-Catenin Protein-Protein Interactions

(A) Potential interactions between β-catenin helix C and ICAT. β-catenin-R1C (βcat-R1C, cyan) is superimposed onto β-catenin/ICAT (βcat/ICAT, yellow and magenta), based on the Cα’s of β-catenin armadillo repeats 10–12 (residues 563–663).

(B) Helix C is unlikely to affect the interaction between the phophorylated E-cadherin and β-catenin armadillo repeats. β-catenin-R1C (βcat-R1C, cyane) is superimposed with β-catenin/phospho-Ecadherin (βcat/pEcad, yellow and red, PDB code: 1I7W) based on all twelve armadillo repeats.

(C) The helix C is required for β-catenin to interact with Chibby. Purified MBP-Chibby protein was immobilized to amylose beads; purified β-catenin and β-catenin fragments were tested for their binding to immobilized MBP-Chibby. The total input and bound β-catenin and β-catenin fragments were visualized by Western blot.

ICAT is a physiological inhibitor of Wnt signals that prevents the binding of Tcf to β-catenin (Tago et al., 2000). Previous crystal structures showed that a three-helix bundle of ICAT anchors at the C terminus of the armadillo repeat region (Daniels and Weis, 2002; Graham et al., 2002). Structural superposition reveals that the helix C of β-catenin is in close contact with one of the ICAT α helices (ICAT helix 1 in Figure 4A). Additional interface appears between β-catenin’s helix C residues Lys672, Ser675, Val676, Thr679, Leu682, Phe683, and ICAT’s helix 1 residues Pro11, Glu12, Met14, Tyr15, Gln18, Arg21, Val22 (Figure 4A). The involvement of the helix C in ICAT binding is consistent with the calorimetric analysis result that the β-cat-enin-R1C binds six times tighter than the β-catenin armadillo repeat domain (Choi et al., 2006). However, it is not very straightforward to predict side-chain interactions from the structural superposition. Therefore, a thorough understanding of the β-catenin/ICAT interaction will require future structural studies. In comparison, from the structural superposition, it is clear that helix C of β-catenin is unlikely involved in the binding of E-cadherin or APC. The helix C sits on a hydrophobic surface formed by the last armadillo repeat that is far away from the structural groove where E-cadherin and APC bind (Huber and Weis, 2001; Figure 4B).

To further understand the role of β-catenin helix C in protein interactions, we also tested the binding of Chibby, another nuclear β-catenin-associated antagonist of the Wnt/Wingless pathway (Takemaru et al., 2003). Purified MBP-Chibby protein was immobilized to amylose beads; purified β-catenin and β-catenin fragments were tested for their binding to immobilized MBP-Chibby. Consistent with previous results (Takemaru et al., 2003), the armadillo repeat domain (residues 133–665) of β-catenin interacts with MBP-Chibby very weakly (Figure 4C). In comparison, addition of the 21 residues of helix C to the armadillo repeat domain (β-catenin fragment 138–686) dramatically increased the binding of β-catenin to MBP-Chibby. This result confirms that the helix C plays a critical role in Chibby binding. In addition to Chibby and ICAT, the helix C may be involved in interactions with many other β-catenin binding proteins, in particular those involved in the transactivation of Wnt responsive genes (see Discussion). Future studies will be needed to test how the helix C may interact with Wnt pathway transcriptional coactivators and inhibitors.

β-Catenin Terminal Domains Do Not Form Stable Complex with the Armadillo Repeat Domain in Trans

Our crystal structures demonstrate that the N- and C-terminal domains of β-catenin are structurally flexible and clearly do not interact with the armadillo repeat domain with a static conformation. To test whether the absence of stable intramolecular interactions in the crystal structure is caused by crystal packing artifacts, we first conducted an in vitro binding assay between the β-catenin armadillo repeat region and its terminal domains. The result shows that the GST-armadillo repeat region (GST-arm) does not pull down significant amounts of N-terminal domain (N) or C-terminal domain (C) compared to the result from control experiments using GST, even when excess terminal domain proteins were used (Figure S1, see the Supplemental Data available with this article online). These results are consistent with our crystal structures, suggesting that the intramolecular interactions in the native β-catenin molecule structures, if existent, are weak and transient.

NMR Studies Suggest a Highly Dynamic Intramolecular Interaction in β-Catenin

To further study the potential intramolecular interactions in the full-length β-catenin, we used the more sensitive NMR titration technique to explore any perturbations that the unlabeled proteins of β-catenin armadillo repeat region could cause when added to N-labeled β-catenin C-terminal tail (residues 687–781, not including the helix C). Consistent with our X-ray crystallography studies, the N-TROSY spectrum of N-labeled β-catenin C-terminal tail alone shows poor dispersion of the resonance, indicating that the C-terminal tail per se does not have any folded structure (Figure 5). When the unlabeled β-catenin(1–686) fragment was added into the solution, except some minor changes indicated by the arrows, no significant chemical shift perturbation could be observed (Figure 5). Clearly, the β-catenin C-terminal tail remains unstructured in the presence of millimolar N-terminal and armadillo repeat domains in trans.

An external file that holds a picture, illustration, etc.
Object name is nihms98359f5.jpg
NMR Analysis of Potential Intramolecular Interactions of β-Catenin

NMR studies suggest highly dynamic interactions between the C-terminal tail and the β-catenin N-R12 fragment N-TROSY spectra of the N-labeled β-catenin(687–781) alone (green) and with unlabeled β-catenine(1–686) (red). Poor dispersion of the resonance indicates that β-catenin(687–781) is unstructured in the absence or presence of β-catenine(1–686). Some minor chemical shift perturbations, as indicated by the arrows, suggest that β-catenin(687–781) may interact with β-catenine(1–686) in a highly dynamic manner.

The N-terminal domain of β-catenin has a poor expression yield and tends to be degraded during purification, which prevented us from carrying out similar NMR experiment with the N-terminal domain. The surface of the armadillo repeat domain of β-catenin, in particular the ligand-binding structural groove, is highly positively charged (Huber et al., 1997). In contrast, the C-terminal tail is highly negatively charged (with an estimated PI of ~4). It is possible that those minor chemical shift perturbations are due to the nonspecific charge-charge interactions between β-catenin armadillo repeat domain and the C-terminal tail. Such small perturbations and a lack of NMR peak broadening expected for stable interactions also indicate that that the potential interaction between the armadillo repeat and C-terminal domain is highly dynamic and variable.

Crystal Structure of β-Catenin-R1C

We first crystallized a human β-catenin fragment (termed as β-catenin-R1C, residues 138–781), which contains both armadillo repeat domain (residues 138–664) of β-catenin and β-catenin C-terminal domain (residues 665–781). Crystals of this β-catenin-R1C fragment contain one molecule per asymmetric unit. The structure was first solved by molecular replacement (MR). In addition to the armadillo repeat domain that is used as the search model, we instantly observed clear electron density for the first 27 residues (residues 665–691) of the β-catenin C-terminal domain, which consist of an α helix (667–683, referred to as helix C) and a loop with a 310-helical turn (684–691). But the rest of the C-terminal domain (referred to as the C-terminal tail in this work) was invisible in the electron density map from MR. To avoid model bias intrinsic to the MR method, we also determined the structure de novo by the SIRAS method and refined the structure to 2.2 Å (Table 1). The initial experimental map has easily recognizable features with excellent contrast and continuity within the protein region (Figure 1A). Nevertheless, only residues 665–691 of β-catenin C-terminal domain, which were also identified by MR method, were visible in the electron density maps before and after density modification. Further structure refinement did not reveal any more electron density belonging to the β-catenin C-terminal tail. The crystal has been proven to contain intact β-catenin-R1C fragment by the SDS electrophoresis of the dissolved crystals (Figure 1C). These data demonstrate that the C-terminal tail is flexible in the β-catenin-R1C structure.

An external file that holds a picture, illustration, etc.
Object name is nihms98359f1.jpg
Human β-Catenin-R1C Crystal Structure

(A) Stereoview of SIRAS experimental map of β-catenin-R1C C-terminal domain helix contoured at 1 σ.

(B) β-catenin-R1C crystal structure. Each armadillo repeat of β-catenin, except repeat 7, is composed of three helices that are shown as blue (helix 1), green (helix 2), and yellow (helix 3) cylinders, whereas the C-terminal domain is colored in red. The C-terminal domain α helix is named helix C.

(C) SDS-gel analysis of β-catenin-R1C crystals. Lane 1, molecular weight marker. Lane 2, purified β-catenin armadillo repeat domain protein (residues 133–665). Lane 3, purified β-catenin-R1C protein (residues 133–781). Lane 4, dissolved β-catenin-R1C crystals.

Table 1

Structure Determination and Refinement of Human β-Catenin-R1C and Full-Length Zebrafish β-Catenin

Data Collectionβ-Catenin-R1CFull-Length β-Catenin
Data setsMecury derivativeNativeNative
Space groupI422I422P212121
Cell Dimensions
a, b, c (Å)106.483, 106.483, 259.830104.229, 104.229, 259.39171.623, 112.819, 123.091
Resolution (Å)a50.0–2.5 (2.59–2.50)50.0–2.2 (2.28–2.20)30.0–3.4 (3.52–3.40)
Rmerg (%)a8.2 (53.8)7.5 (61.1)9.1 (49.6)
I/σ(I)a24.6 (2.7)21.8 (1.9)12.7 (2.2)
Completeness (%)a99.7 (99.0)96.8 (81.2)89.1 (90.3)
Redundancy9.18.53.4
Mecury sites found2
Refinement Statistics
Resolution (Å)50.0–2.230.0–3.4
Number of reflections2978311140
Rwork/Rfree (%)21.3/24.928.1/31.9
Number of Atoms
 Protein40324188
 Water280
B Factors
 Protein40.166.3
 Water34.1
Rms Deviations
 Bond length (Å)0.0060.003
 Bond angle (°)1.10.7
In parentheses are corresponding numbers for the highest resolution shell.

The α helix of β-catenin C-terminal domain (helix C) packs against otherwise exposed hydrophobic residues from the last armadillo repeat, capping the C-terminal end of the armadillo repeat region. The helix C is connected to and runs parallel to the helix 3 of armadillo repeat 12 (Figure 1B). A series of hydrophobic side chains from the helix C, Leu674, Leu678, and Leu682, fit into a hydrophobic patch formed by the second and third helices of armadillo repeat 12, coordinating tightly with Leu644, Ala652, Ala656, and Phe660 (Figures 1A, 2A, and 2C). Side chain nitrogens of Lys671 and Arg684 at each end of the helix C attach to armadillo repeats through hydrogen bonds with backbone carbonyl oxygens of Glu664 and Ser646, respectively (Figures 2A and 2C). In comparison, the other surface of the helix C is composed of hydrophilic residues exposed to solvent (Figures 2A and 2C). It is remarkable that the helix C, including the three leucine residues that dock the helix C onto the armadillo repeat domain, is highly conserved in sequence, while the rest of the C-terminal domain has a divergent sequence (Figure 2B). These observations strongly suggest that this capping helix C is a shared structural feature in the β-catenin family.

An external file that holds a picture, illustration, etc.
Object name is nihms98359f2.jpg
Interactions between β-Catenin C-Terminal Domain and Central Armadillo Repeat Domain

(A) Electrostatic surface of the β-catenin armadillo repeat domain covered by the β-catenin C-terminal domain. Residues from the armadillo repeat domain are labeled in black. β-catenin C-terminal domain is in a stick model labeled in red. Key residues of the C-terminal domain interacting with the armadillo repeat domain are highlighted in green for hydrophobic interactions and yellow for charge-charge interactions.

(B) Sequence alignment of the β-catenin C-terminal domain with CLUSTALW. Invariant residues are labeled with stars in red, strongly similar residues are labeled with (:) in green, and weakly similar residues are labeled with (.) in blue. Sequences of the C-terminal domain visible in the β-catenin-R1C crystal structure are framed. The conserved hydrophobic residues (L674, L678, and L682) that docked the helix C into the armadillo repeat domain are marked with arrows.

(C) Ligplot presentation of interactions between β-catenin residues in the C-terminal domain (purple) and the armadillo repeat domain (red). Hydrogen bonds and charge-charge interactions are designated with green dashed lines and distance in Å. A red starburst together with a red dashed line represents hydrophobic interactions.

Crystal Structure of Full-Length β-Catenin

To provide a more complete picture of β-catenin structure for understanding the mechanisms of Wnt signaling and cell adhesion, we extensively screened the crystallization conditions of full-length β-catenin from various organisms. The full-length zebrafish β-catenin was the only β-catenin that yielded single crystals. The structure of the full-length zebrafish β-catenin has been solved and refined with numerous data sets collected from different crystals (Figure 3A), with the highest resolution being 3.4 Å (Table 1). The electron density map from molecular replacement has been carefully searched. No clear electron density was observed for the two terminal domains except for the helix C in the C-terminal domain and an additional α helix flanking the N terminus of armadillo repeat 1. A shorter N-terminal α helix has been observed in the crystal structures of β-catenin in complex with Tcf-3 or the unphosphorylated E-cadherin cytoplasmic domain complex (Graham et al., 2000; Huber and Weis, 2001). The final crystal structure obtained from full-length zebrafish β-catenin protein crystals contains residues 126–681 (corresponding to human β-catenin residues 127–682; Figure 3A). For simplicity, all β-catenin residue numbers used in this work are these of corresponding human β-catenin unless specified. The presence of full-length β-catenin in the crystal has been clearly verified by the SDS electrophoresis of the washed and dissolved single crystals (Figure 3B).

An external file that holds a picture, illustration, etc.
Object name is nihms98359f3.jpg
Full-Length Zebrafish β-Catenin Crystal Structure

(A) Crystal structure of full-length zebrafish β-catenin. The color scheme of β-catenin armadillo repeats is the same as that in Figure 1. The C-terminal helix (helix C) is colored in red. The N-terminal helix is colored in blue and green with an arrow pointing to the kink. All β-catenin residue numbers shown in this figure are these of corresponding residue numbers of human β-catenin.

(B) SDS-gel analysis of the dissolved full-length zebrafish β-catenin crystals. Lane 1, molecular weight marker. Lane 2, dissolved crystals. Lane 3, purified full-length zebrafish β-catenin.

(C) Structural superposition between human β-catenin-R1C (red) and full-length zebrafish β-catenin (cyan) around helix C.

It is remarkable that the helix C was observed in both β-cate-nin-R1C and full-length β-catenin crystal structures. The helix C is unlikely a crystal packing artifact, since crystal packing environments in these two crystal structures are very different and helix C is mostly not involved in direct crystal packing. The helix C together with the armadillo repeats 10–12 from β-catenin-R1C and full-length β-catenin structures are closely superimopsable, with the Cα rmsd of the helix C (residues 667–682 in human β-catenin-R1C) being 1.26 Å (Figure 3C).

The Helix C Plays Important Roles in β-Catenin-Ligand Interactions

β-catenin participates in a variety of protein-protein networks in both Wnt-signaling pathway and cell-adherens junctions. In published β-catenin complex crystal structures, the armadillo repeat 12 of β-catenin forms part of the binding interfaces for APC, E-cadherin, and ICAT (Daniels and Weis, 2002; Graham et al., 2002; Ha et al., 2004; Huber and Weis, 2001; Xing et al., 2004). To investigate whether these β-catenin partners interact with the helix C, we superimposed the crystal structure of β-cat-enin-R1C with those of the β-catenin/ICAT, the β-catenin/APC, and the β-catenin/E-cadherin complexes (Figures 4A and 4B).

An external file that holds a picture, illustration, etc.
Object name is nihms98359f4.jpg
The Role of Helix C in β-Catenin Protein-Protein Interactions

(A) Potential interactions between β-catenin helix C and ICAT. β-catenin-R1C (βcat-R1C, cyan) is superimposed onto β-catenin/ICAT (βcat/ICAT, yellow and magenta), based on the Cα’s of β-catenin armadillo repeats 10–12 (residues 563–663).

(B) Helix C is unlikely to affect the interaction between the phophorylated E-cadherin and β-catenin armadillo repeats. β-catenin-R1C (βcat-R1C, cyane) is superimposed with β-catenin/phospho-Ecadherin (βcat/pEcad, yellow and red, PDB code: 1I7W) based on all twelve armadillo repeats.

(C) The helix C is required for β-catenin to interact with Chibby. Purified MBP-Chibby protein was immobilized to amylose beads; purified β-catenin and β-catenin fragments were tested for their binding to immobilized MBP-Chibby. The total input and bound β-catenin and β-catenin fragments were visualized by Western blot.

ICAT is a physiological inhibitor of Wnt signals that prevents the binding of Tcf to β-catenin (Tago et al., 2000). Previous crystal structures showed that a three-helix bundle of ICAT anchors at the C terminus of the armadillo repeat region (Daniels and Weis, 2002; Graham et al., 2002). Structural superposition reveals that the helix C of β-catenin is in close contact with one of the ICAT α helices (ICAT helix 1 in Figure 4A). Additional interface appears between β-catenin’s helix C residues Lys672, Ser675, Val676, Thr679, Leu682, Phe683, and ICAT’s helix 1 residues Pro11, Glu12, Met14, Tyr15, Gln18, Arg21, Val22 (Figure 4A). The involvement of the helix C in ICAT binding is consistent with the calorimetric analysis result that the β-cat-enin-R1C binds six times tighter than the β-catenin armadillo repeat domain (Choi et al., 2006). However, it is not very straightforward to predict side-chain interactions from the structural superposition. Therefore, a thorough understanding of the β-catenin/ICAT interaction will require future structural studies. In comparison, from the structural superposition, it is clear that helix C of β-catenin is unlikely involved in the binding of E-cadherin or APC. The helix C sits on a hydrophobic surface formed by the last armadillo repeat that is far away from the structural groove where E-cadherin and APC bind (Huber and Weis, 2001; Figure 4B).

To further understand the role of β-catenin helix C in protein interactions, we also tested the binding of Chibby, another nuclear β-catenin-associated antagonist of the Wnt/Wingless pathway (Takemaru et al., 2003). Purified MBP-Chibby protein was immobilized to amylose beads; purified β-catenin and β-catenin fragments were tested for their binding to immobilized MBP-Chibby. Consistent with previous results (Takemaru et al., 2003), the armadillo repeat domain (residues 133–665) of β-catenin interacts with MBP-Chibby very weakly (Figure 4C). In comparison, addition of the 21 residues of helix C to the armadillo repeat domain (β-catenin fragment 138–686) dramatically increased the binding of β-catenin to MBP-Chibby. This result confirms that the helix C plays a critical role in Chibby binding. In addition to Chibby and ICAT, the helix C may be involved in interactions with many other β-catenin binding proteins, in particular those involved in the transactivation of Wnt responsive genes (see Discussion). Future studies will be needed to test how the helix C may interact with Wnt pathway transcriptional coactivators and inhibitors.

β-Catenin Terminal Domains Do Not Form Stable Complex with the Armadillo Repeat Domain in Trans

Our crystal structures demonstrate that the N- and C-terminal domains of β-catenin are structurally flexible and clearly do not interact with the armadillo repeat domain with a static conformation. To test whether the absence of stable intramolecular interactions in the crystal structure is caused by crystal packing artifacts, we first conducted an in vitro binding assay between the β-catenin armadillo repeat region and its terminal domains. The result shows that the GST-armadillo repeat region (GST-arm) does not pull down significant amounts of N-terminal domain (N) or C-terminal domain (C) compared to the result from control experiments using GST, even when excess terminal domain proteins were used (Figure S1, see the Supplemental Data available with this article online). These results are consistent with our crystal structures, suggesting that the intramolecular interactions in the native β-catenin molecule structures, if existent, are weak and transient.

NMR Studies Suggest a Highly Dynamic Intramolecular Interaction in β-Catenin

To further study the potential intramolecular interactions in the full-length β-catenin, we used the more sensitive NMR titration technique to explore any perturbations that the unlabeled proteins of β-catenin armadillo repeat region could cause when added to N-labeled β-catenin C-terminal tail (residues 687–781, not including the helix C). Consistent with our X-ray crystallography studies, the N-TROSY spectrum of N-labeled β-catenin C-terminal tail alone shows poor dispersion of the resonance, indicating that the C-terminal tail per se does not have any folded structure (Figure 5). When the unlabeled β-catenin(1–686) fragment was added into the solution, except some minor changes indicated by the arrows, no significant chemical shift perturbation could be observed (Figure 5). Clearly, the β-catenin C-terminal tail remains unstructured in the presence of millimolar N-terminal and armadillo repeat domains in trans.

An external file that holds a picture, illustration, etc.
Object name is nihms98359f5.jpg
NMR Analysis of Potential Intramolecular Interactions of β-Catenin

NMR studies suggest highly dynamic interactions between the C-terminal tail and the β-catenin N-R12 fragment N-TROSY spectra of the N-labeled β-catenin(687–781) alone (green) and with unlabeled β-catenine(1–686) (red). Poor dispersion of the resonance indicates that β-catenin(687–781) is unstructured in the absence or presence of β-catenine(1–686). Some minor chemical shift perturbations, as indicated by the arrows, suggest that β-catenin(687–781) may interact with β-catenine(1–686) in a highly dynamic manner.

The N-terminal domain of β-catenin has a poor expression yield and tends to be degraded during purification, which prevented us from carrying out similar NMR experiment with the N-terminal domain. The surface of the armadillo repeat domain of β-catenin, in particular the ligand-binding structural groove, is highly positively charged (Huber et al., 1997). In contrast, the C-terminal tail is highly negatively charged (with an estimated PI of ~4). It is possible that those minor chemical shift perturbations are due to the nonspecific charge-charge interactions between β-catenin armadillo repeat domain and the C-terminal tail. Such small perturbations and a lack of NMR peak broadening expected for stable interactions also indicate that that the potential interaction between the armadillo repeat and C-terminal domain is highly dynamic and variable.

DISCUSSION

β-catenin plays central roles in both cell adhesion and Wnt signaling, through its interaction with numerous protein partners. A complete understanding of β-catenin structure is not only critical for understanding β-catenin-partner interactions in cell regulation but is also essential for development of small-molecule inhibitors to terminate canonical Wnt signaling pathway, which may be useful for cancer treatments. Here, we report two distinct crystal forms in an effort to unravel the three-dimensional structure of β-catenin. In both crystal forms, in addition to the armadillo repeat domain, we observed an α helix, the helix C, at the beginning of the β-catenin C-terminal domain. The conformation and orientation of the helix C are essentially the same in both crystal structures as shown by our structural superposition (Figure 3C), which confirms that the formation of helix C is independent of specific crystal packing. It should be noted that it was already recognized some time ago that high level sequence similarity continued beyond what are now known to be the armadillo repeats into the C-terminal domain, a region that turns out to be the helix C. In fact, in the original delineation of the “armadillo repeats” of Armadillo, the authors included a thirteenth armadillo repeat that encompasses this helix in their predictions (Peifer and Wieschaus, 1990). In our crystal structures, the helix C runs in parallel to helix 3 of armadillo repeat 12 (Figures 1B and and3A).3A). Based on the relative positions of the three helices in each armadillo repeat, the helix C should not be considered as the beginning of armadillo repeat 13. Instead, it caps the hydrophobic surface formed by the C-terminal end of the armadillo repeats. Therefore, the helix C, a new, to our knowledge, structure element identified in this work, forms part of the superhelical structure core of β-catenin together with the armadillo repeat domain.

Previous work in the Drosophila system has suggested that the C-terminal domain is essential for the role of β-catenin in transcriptional regulation, which can be subdivided into regions of essential importance and regions that are less important. These are delineated by two truncation mutations in Drosophila Armadillo, armXM19 and armH8.6. The former is null for Wnt signaling, while the latter is reduced but not null (Cox et al., 1999). Strikingly, the truncation in armXM19 ends almost precisely at the end of the Arm repeats, while the truncation in armH8.6 is placed to almost precisely leave the C helix intact and delete the rest of the C-terminal domain. Therefore our crystal structure correlates well with previous genetic studies and provides a mechanistic basis for understanding these functional observations.

Several transcriptional coactivators are known to interact with the C-terminal domain. It has been proposed that these β-catenin binding partners interact with β-catenin and function in different stages of transcription (Stadeli et al., 2006; Willert and Jones, 2006). Since the interactions of β-catenin with those transcriptional coactivators appear conserved during evolution, and the helix C is a highly conserved region of the C-terminal domain, it is likely that the helix C plays an important role in the recruitment of these transcriptional coactivators and in the subsequent activation of Wnt-responsive genes. It should be noted that the transcription of Wnt-responsive genes is the last step of canonical Wnt signaling, and constitutive transactivation of Wnt-responsive genes, such as c-myc and cyclin D1, is the hallmark of many cancers. Therefore, the structural interface of β-catenin and these β-catenin binding partners is an important drug targets for cancer treatment. The revelation of helix C provides key insights into these β-catenin-coactivator interactions with implications for future structure-based drug design.

Given the important role of these β-catenin-partner interactions in Wnt signaling, it is not surprising that the same β-catenin region is also the binding site of transcriptional inhibitors such as ICAT and Chibby (Tago et al., 2000; Takemaru et al., 2003). Both ICAT and Chibby compete with Tcf for β-catenin binding. In addition, it has been shown that ICAT competes directly with CBP/p300 for β-catenin binding (Daniels and Weis, 2002). Consistent with our hypothesis that the helix C is a key player in Wnt-stimulated transcription, β-catenin-Chibby interaction requires the helix C, and ICAT may also interact extensively with the helix C when it binds to β-catenin (Figure 4A). Other than these transcriptional coactivators/inhibitors that interact with the β-catenin R11-C region, Tcf, APC, and E-cadherin have essentially the same binding site in the structural groove formed by the armadillo repeats (Eklof Spink et al., 2001; Graham et al., 2000; Ha et al., 2004; Huber and Weis, 2001; Poy et al., 2001; Xing et al., 2004). They may not directly interact with the helix C according to our superimposed crystal structures (Figure 4B). The possibility that the helix C is important for the transactivation of Wnt-responsive genes, but not the turnover of β-catenin through APC and cell adhesion through cadherins, makes the helix C region an even more interesting target for drug design. It is unclear whether the binding of any β-catenin partner would induce significant conformational change in the C helix. Future biochemical and structural studies of β-catenin with transcriptional coactivators/inhibitors will be important for designing drugs that will disrupt β-catenin-coactivator but not β-catenin-inhibitor interactions.

In the N terminus, the first armadillo repeat varies from other armadillo repeats in that its helix 1 and helix 2 merge into one elongated and kinked α helix. Comparison of our full-length β-catenin structure with other β-catenin structures with a visible first armadillo repeat (Graham et al., 2000; Huber and Weis, 2001) reveals a significant hinge motion around Arg151 in the kinked α helix of armadillo repeat 1 (Figure S2). Part of E-cadherin associates with residues C-terminal to L148 in this α helix (Huber and Weis, 2001). The α-catenin dimerization domain interacts with β-catenin residues 118–149, and it was proposed that the melting of the helical conformation near β-catenin residues 141–149 is required for α-catenin binding (Pokutta and Weis, 2000). The observed hinge motion in this kinked α helix is consistent with the notion that the N-terminal end of β-catenin armadillo repeat domain (residues 118–149) is structurally dynamic, which may allow for efficient α-catenin binding.

It was previously proposed that there may be intramolecular interactions among N- and C-terminal domains with the armadillo repeat domain, which regulates β-catenin protein interactions (Castano et al., 2002; Cox et al., 1999; Piedra et al., 2001). In contrast, quantitative isothermal calorimetry (ITC) study has shown that the N- and C-terminal domains did not affect the binding affinity of β-catenin for tight β-catenin binding partners. In our crystal structures, the N- and the C-terminal tails are disordered despite the presence of intact β-catenin R1C and full-length β-catenin in respective crystals. Therefore, we conclude that the N- and C-terminal tails do not interact with the armadillo repeat domain with a specific conformation.

Our NMR studies also directly demonstrate that the β-catenin C-terminal tail is unstructured in solution, in the presence or absence of the N-terminal and armadillo repeat domains (β-catenin N-R12). However, the presence of β-catenin N-R12 does induce minor NMR spectra perturbation, which may be explained by highly dynamic, nonspecific interactions between the C-terminal tail and the β-catenin N-R12 fragment. For technical reasons, the interaction between the N-terminal tail and the β-catenin R1-C fragment was not studied by NMR. Although the potential intermolecular interaction between the C-terminal domain and the rest of β-catenin (N-R12) is weak and transient, this interaction may become significant in the context of intramolecular interactions, since the local concentration of the C-terminal tail in the full-length β-catenin is much higher than what was used in the titration experiment. Because both β-catenin N- and C-terminal tails are highly negatively charged, and the structural groove of armadillo repeat domain is highly positively charged (Huber et al., 1997), it is predictable that both N- and C-terminal tails of β-catenin interact with the structural groove of armadillo repeat domain through nonspecific charge-charge interactions. This hypothesis is consistent with the results of thermodynamic studies that β-catenin terminal domains do affect the binding of weak groove binders (with Kd in low- to sub-μM range) such as Axin and unphosphorylated APC, while no effect was observed for strong binders (with Kd in low nM range; Choi et al., 2006).

Based on all these structural and thermodynamic data, we suggest that the N- and C-terminal tails may prevent nonspecific protein-protein interactions of the armadillo repeat domain, by forming a highly dynamic “shield” around the armadillo repeat domain. In some sense, the N- and the C-terminal tails are “intra-molecular chaperones” of the armadillo repeat domain, which increase the binding specificity and prevent the self-aggregation of the armadillo repeat domain region. In fact, the full-length β-catenin molecules are highly soluble in solution, while both the armadillo repeat domain and the armadillo repeat domain plus the helix C (arm-C) region have a strong tendency to aggregate in solution (data not shown).

Finally, it should be noted that all our structural and previously published ITC data were obtained with unmodified β-catenin proteins or fragments produced in bacteria. In the cell, it has been demonstrated that a Wnt signal results in a form of β-catenin that binds selectively with Tcf but not cadherin (Gottardi and Gumbiner, 2004). Another form of β-catenin that binds cadherin has a more accessible C-terminal domain (Gottardi and Gumbiner, 2004). It suggests the presence of different molecular conformations of β-catenin, which is regulated by Wnt signals. Therefore, our results demonstrate that, in unmodified β-catenin molecules, the N- and the C-terminal tails interact with the armadillo repeat domain in a highly dynamic and variable manner. However, posttranslational modifications on β-catenin, such as those resulted from a Wnt signal that remains to be identified, may stabilize the N- and/or C-terminal tails in a more specific conformation and directly regulate the binding of specific β-catenin partners.

EXPERIMENTAL PROCEDURES

Protein Purification and Crystallization

The full-length zebrafish β-catenin (residues 1–780) and human β-catenin-R1C (residues 138–781) proteins used in this work were overexpressed by using pET28a vector in E. coli BL21(DE3)Star cells. Both proteins contain a TEV protease cleavage site between the His6-tag and β-catenin. Cells were lysed, and the supernatant was loaded to the affinity column containing Ni-NTA beads for His-tagged protein (QIAGEN). Eluents from the affinity column were incubated with TEV at room temperature overnight. β-catenin protein in the cleavage solution was further purified first by ion exchange with Q-Sepharose HP (Amersham Pharmacia) and then with a Superdex 200 size exclusion column (Amersham Pharmacia). Purified proteins were concentrated to ~5–10 mg/ml in 20 mM Tris (pH 8.0), 50 mM NaCl, 2 mM DTT. Protein aliquots were frozen with liquid nitrogen and stored in −80°C freezer.

The optimized condition for the crystallization of β-catenin-R1C contained 1 μl β-catenin-R1C (6 mg/ml in 20 mM Tris [pH 8.0], 50 mM NaCl, 2 mM DTT) and 1 μl reservoir solution (0.1 M sodium citrate [pH 5.6], 0.4 M NaCl, and 7% Jeffamine ED-2001). Crystals grew to 0.2 × 0.2 × 0.05 mm after one month, which were then soaked stepwise into soaking solution: 80 mM Na Citrate [pH 5.6], 20 mM Tris [pH 8.5], 0.4 M NaCl, 7% Jeffamine ED-2001, 22% glycerol, 10% PEG 8000, and 5 mM DTT, before data collection. Various heavy-atom compounds were added to β-catenin-R1C soaking drops to search for the isomorphous derivatives. A mercury derivative was successfully obtained when crystals were soaked in Methylmercury(II) Chloride (Hampton) overnight at room temperature. The best crystals of full-length zebrafish β-catenin were obtained by mixing 1 μl of protein solution with 1 μl of reservoir solution (0.1 M Tris [pH 7.5], 10% ethanol, 25 mM spermine tetrahydrochlor-ide). Crystals were stepwise cryoprotected with 25% glycerol.

Data Collection and Structure Determination

Data sets of human β-catenin-R1C and its mercury derivative and full-length zebrafish β-catenin were collected at the Advanced Photon Source (APS) beamline 19ID. Data sets were processed and scaled with the HKL2000 program suite (Otwinowski and Minor, 1997). A 2.2 Å data set of native β-catenin-R1C and a 2.5 Å data set of mercury derivative were used in the structure determination. Initial phases for the β-catenin-R1C structure were calculated independently by AMoRe (Navaza, 1994) by using the molecular replacement technique and by SOLVE (Terwilliger and Berendzen, 1999) by using the SIRAS phasing technique and improved with density modification in RESOLVE (Terwilliger, 2000). To improve the obscure electron density in the solvent region potentially belonging to the C tail, a variety of experimental maps were generated with different schemes of density modifications. The initial map was fitted with the armadillo repeat domain structure of murine β-catenin (PDB code: 3BCT). Xtalview (McRee, 1999) and CNS (Brunger et al., 1998) were used for further model building and refinement. Throughout the refinement process, 8% of the reflections of the 2.2 Å data set were excluded as a test set, and the data were subject to a 1σ cutoff. During model building, both 2Fo-Fc maps and difference (Fo-Fc) maps were calculated to locate the missing portion, and composite simulated annealing omit maps were calculated to reduce the effects of model bias. Geometry of the final model was checked with PROCHECK. The final model of β-catenin-R1C contains residues 149–691. 93.1% and 0.0% residues in the β-catenin-R1C structure are in the core and disallowed regions of Ramachandran plot, respectively.

The structure of full-length zebrafish β-catenin was solved by molecular replacement with AMoRe. The best solution was obtained by using the β-catenin structure in the β-catenin/E-cadherin complex (PDB code: 1I7X) as the search model. The model was refined with CNS programs (Brunger et al., 1998), and model building was conducted in Xtalview (McRee, 1999). Structure determination and refinement was done with multiple data sets collected from full-length β-catenin crystals to confirm the unknown structure of the terminal domains. The statistics with the highest resolution (3.4 Å) data set were listed in the Table 1. A test set of 5.0% of the reflections and a 1σ cutoff of the data was used in the refinement. 90.3% residues of the final model are in the core region of Ramachandran plot, and none is in the disallowed region.

In Vitro β-Catenin-Chibby Binding Assay

One microgram of MBP or MBP-Chibby was incubated at 4°C for 30 min with 3 μl of amylose resin (New England Biolabs, MA), which was preequilibrated in 10 μl of protein-binding buffer (PBB; 20 mM Tris [pH 7.4], 0.2 M NaCl, 1 mM EDTA, 10 mM 2-mercaptoethanol). After washing with PBB, immobilized MBP or MBP-Chibby was incubated at 4°C for 1 hr in 10 μl of PBB, with 2 μg of purified human FL-β-catenin, β-catenin R1-C (residues 134–781), β-catenin armadillo repeat domain (arm; residues 133–665), β-catenin armadillo repeat domain with the helix C (arm-C; residues 138–686), respectively. The beads were collected and washed with PBB and then boiled in SDS loading buffer for SDS-PAGE electrophoresis. Since MBP-Chibby migrates to the same position as β-catenin armadillo repeat domain (arm) in SDS-PAGE gels, FL-β-catenin, β-catenin R1-C, arm, or arm-C bound to MBP or MBP-Chibby were visualized by western blot with an antibody against the β-catenin armadillo repeat domain (Calbiochem, CA; Figure 4C).

In Vitro Pull-Down of the β-Catenin Armadillo Repeat Domain by N- and C-Terminal Domains

His-tagged N-terminal (residues 1–137) and C-terminal (residues 666–781) domains of human β-catenin and GST-tagged armadillo repeat domain of β-catenin (GST-arm) were expressed in E. coli BL21(DE3)Star cells, and purified with affinity column followed by the cleavage of the affinity tag with TEV protease and subsequent Q-sepharose HP column. GST-arm (or GST, ~0.2 nmol) were incubated with N- and C-terminal domains at molar ratios 1:1 and 5:1 (terminal domains versus GST protein) for 30 min at 20°C. Incubations were performed in binding buffer: 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, and 0.1% (w/v) Triton X-100 in a final volume of 200 μl. Protein complexes were isolated by incubation with 40 μl of a 50% (w/v) suspension of glutathione-Sepharose 4B in the binding buffer for 30 min at 20°C. Beads were collected via microcentrifugation and washed three times with binding buffer. Samples were separated and analyzed by SDS-polyacrylamide gel electrophoresis.

NMR Titration Experiment

A GST-tagged β-catenin fragment (residues 1–686, with a TEV site following GST-tag) containing N-terminal domain, armadillo repeats, and the C-terminal α helix (helix C) was expressed and purified in an essentially similar way as other β-catenin proteins described. Purified β-catenin(1–686) was concentrated to 0.07 mM in 20 mM Tris [pH 7.5], 0.15 M NaCl, and 2 mM DTT buffer. A His-tagged β-catenin fragment (residues 687–781, with a TEV site following His-tag) covering the rest of the C-terminal domain was expressed by E. coli BL21(DE3)Star cells in a supplemented minimal media with NH4Cl as the sole nitrogen source. N labeled β-catenin (687–781) was purified and concentrated to 0.5 mM in 20 mM Tris (pH 7.5), 0.1 M NaCl, and 2 mM DTT.

NMR data were recorded with a Bruker Avance 800 MHz 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. N-TROSY experiments were performed to detect chemical shift perturbation in the spectra of N-labeled β-catenin(687–781) with different molar ratios (from 1:0 to 1:2) of unlabeled β-catenin(1–686) indicative of possible intermolecular binding. The data were processed and displaced by using Bruker software XWINNMR.

Protein Purification and Crystallization

The full-length zebrafish β-catenin (residues 1–780) and human β-catenin-R1C (residues 138–781) proteins used in this work were overexpressed by using pET28a vector in E. coli BL21(DE3)Star cells. Both proteins contain a TEV protease cleavage site between the His6-tag and β-catenin. Cells were lysed, and the supernatant was loaded to the affinity column containing Ni-NTA beads for His-tagged protein (QIAGEN). Eluents from the affinity column were incubated with TEV at room temperature overnight. β-catenin protein in the cleavage solution was further purified first by ion exchange with Q-Sepharose HP (Amersham Pharmacia) and then with a Superdex 200 size exclusion column (Amersham Pharmacia). Purified proteins were concentrated to ~5–10 mg/ml in 20 mM Tris (pH 8.0), 50 mM NaCl, 2 mM DTT. Protein aliquots were frozen with liquid nitrogen and stored in −80°C freezer.

The optimized condition for the crystallization of β-catenin-R1C contained 1 μl β-catenin-R1C (6 mg/ml in 20 mM Tris [pH 8.0], 50 mM NaCl, 2 mM DTT) and 1 μl reservoir solution (0.1 M sodium citrate [pH 5.6], 0.4 M NaCl, and 7% Jeffamine ED-2001). Crystals grew to 0.2 × 0.2 × 0.05 mm after one month, which were then soaked stepwise into soaking solution: 80 mM Na Citrate [pH 5.6], 20 mM Tris [pH 8.5], 0.4 M NaCl, 7% Jeffamine ED-2001, 22% glycerol, 10% PEG 8000, and 5 mM DTT, before data collection. Various heavy-atom compounds were added to β-catenin-R1C soaking drops to search for the isomorphous derivatives. A mercury derivative was successfully obtained when crystals were soaked in Methylmercury(II) Chloride (Hampton) overnight at room temperature. The best crystals of full-length zebrafish β-catenin were obtained by mixing 1 μl of protein solution with 1 μl of reservoir solution (0.1 M Tris [pH 7.5], 10% ethanol, 25 mM spermine tetrahydrochlor-ide). Crystals were stepwise cryoprotected with 25% glycerol.

Data Collection and Structure Determination

Data sets of human β-catenin-R1C and its mercury derivative and full-length zebrafish β-catenin were collected at the Advanced Photon Source (APS) beamline 19ID. Data sets were processed and scaled with the HKL2000 program suite (Otwinowski and Minor, 1997). A 2.2 Å data set of native β-catenin-R1C and a 2.5 Å data set of mercury derivative were used in the structure determination. Initial phases for the β-catenin-R1C structure were calculated independently by AMoRe (Navaza, 1994) by using the molecular replacement technique and by SOLVE (Terwilliger and Berendzen, 1999) by using the SIRAS phasing technique and improved with density modification in RESOLVE (Terwilliger, 2000). To improve the obscure electron density in the solvent region potentially belonging to the C tail, a variety of experimental maps were generated with different schemes of density modifications. The initial map was fitted with the armadillo repeat domain structure of murine β-catenin (PDB code: 3BCT). Xtalview (McRee, 1999) and CNS (Brunger et al., 1998) were used for further model building and refinement. Throughout the refinement process, 8% of the reflections of the 2.2 Å data set were excluded as a test set, and the data were subject to a 1σ cutoff. During model building, both 2Fo-Fc maps and difference (Fo-Fc) maps were calculated to locate the missing portion, and composite simulated annealing omit maps were calculated to reduce the effects of model bias. Geometry of the final model was checked with PROCHECK. The final model of β-catenin-R1C contains residues 149–691. 93.1% and 0.0% residues in the β-catenin-R1C structure are in the core and disallowed regions of Ramachandran plot, respectively.

The structure of full-length zebrafish β-catenin was solved by molecular replacement with AMoRe. The best solution was obtained by using the β-catenin structure in the β-catenin/E-cadherin complex (PDB code: 1I7X) as the search model. The model was refined with CNS programs (Brunger et al., 1998), and model building was conducted in Xtalview (McRee, 1999). Structure determination and refinement was done with multiple data sets collected from full-length β-catenin crystals to confirm the unknown structure of the terminal domains. The statistics with the highest resolution (3.4 Å) data set were listed in the Table 1. A test set of 5.0% of the reflections and a 1σ cutoff of the data was used in the refinement. 90.3% residues of the final model are in the core region of Ramachandran plot, and none is in the disallowed region.

In Vitro β-Catenin-Chibby Binding Assay

One microgram of MBP or MBP-Chibby was incubated at 4°C for 30 min with 3 μl of amylose resin (New England Biolabs, MA), which was preequilibrated in 10 μl of protein-binding buffer (PBB; 20 mM Tris [pH 7.4], 0.2 M NaCl, 1 mM EDTA, 10 mM 2-mercaptoethanol). After washing with PBB, immobilized MBP or MBP-Chibby was incubated at 4°C for 1 hr in 10 μl of PBB, with 2 μg of purified human FL-β-catenin, β-catenin R1-C (residues 134–781), β-catenin armadillo repeat domain (arm; residues 133–665), β-catenin armadillo repeat domain with the helix C (arm-C; residues 138–686), respectively. The beads were collected and washed with PBB and then boiled in SDS loading buffer for SDS-PAGE electrophoresis. Since MBP-Chibby migrates to the same position as β-catenin armadillo repeat domain (arm) in SDS-PAGE gels, FL-β-catenin, β-catenin R1-C, arm, or arm-C bound to MBP or MBP-Chibby were visualized by western blot with an antibody against the β-catenin armadillo repeat domain (Calbiochem, CA; Figure 4C).

In Vitro Pull-Down of the β-Catenin Armadillo Repeat Domain by N- and C-Terminal Domains

His-tagged N-terminal (residues 1–137) and C-terminal (residues 666–781) domains of human β-catenin and GST-tagged armadillo repeat domain of β-catenin (GST-arm) were expressed in E. coli BL21(DE3)Star cells, and purified with affinity column followed by the cleavage of the affinity tag with TEV protease and subsequent Q-sepharose HP column. GST-arm (or GST, ~0.2 nmol) were incubated with N- and C-terminal domains at molar ratios 1:1 and 5:1 (terminal domains versus GST protein) for 30 min at 20°C. Incubations were performed in binding buffer: 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, and 0.1% (w/v) Triton X-100 in a final volume of 200 μl. Protein complexes were isolated by incubation with 40 μl of a 50% (w/v) suspension of glutathione-Sepharose 4B in the binding buffer for 30 min at 20°C. Beads were collected via microcentrifugation and washed three times with binding buffer. Samples were separated and analyzed by SDS-polyacrylamide gel electrophoresis.

NMR Titration Experiment

A GST-tagged β-catenin fragment (residues 1–686, with a TEV site following GST-tag) containing N-terminal domain, armadillo repeats, and the C-terminal α helix (helix C) was expressed and purified in an essentially similar way as other β-catenin proteins described. Purified β-catenin(1–686) was concentrated to 0.07 mM in 20 mM Tris [pH 7.5], 0.15 M NaCl, and 2 mM DTT buffer. A His-tagged β-catenin fragment (residues 687–781, with a TEV site following His-tag) covering the rest of the C-terminal domain was expressed by E. coli BL21(DE3)Star cells in a supplemented minimal media with NH4Cl as the sole nitrogen source. N labeled β-catenin (687–781) was purified and concentrated to 0.5 mM in 20 mM Tris (pH 7.5), 0.1 M NaCl, and 2 mM DTT.

NMR data were recorded with a Bruker Avance 800 MHz 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. N-TROSY experiments were performed to detect chemical shift perturbation in the spectra of N-labeled β-catenin(687–781) with different molar ratios (from 1:0 to 1:2) of unlabeled β-catenin(1–686) indicative of possible intermolecular binding. The data were processed and displaced by using Bruker software XWINNMR.

Supplementary Material

Sup

Sup

Click here to view.(76K, pdf)

Acknowledgments

We thank the support from the staff at APS beamline 19-ID. This work was supported by National Institutes of Health grant CA 90351 to W.X. R.T.M. is an investigator of the Howard Hughes Medical Institute.

Department of Biological Structure, University of Washington School of Medicine, Seattle, WA 98195, USA
Department of Pharmacology, University of Washington School of Medicine, Seattle, WA 98195, USA
Howard Hughes Medical Institute, University of Washington School of Medicine, Seattle, WA 98195, USA
Institute for Stem Cell and Regenerative Medicine, University of Washington School of Medicine, Seattle, WA 98195, USA
Department of Structural Biology, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA
Correspondence: ude.notgnihsaw.u@uxw
Present address: BCMP, Harvard Medical School, Seeley G. Mudd 130, 250 Longwood Ave., Boston, MA 02115, USA.
Present address: Department of Pharmacological Sciences, State University of New York at Stony Brook, Stony Brook, NY 11794, USA.

SUMMARY

β-catenin plays essential roles in cell adhesion and Wnt signaling, while deregulation of β-catenin is associated with multiple diseases including cancers. Here, we report the crystal structures of full-length zebrafish β-catenin and a human β-catenin fragment that contains both the armadillo repeat and the C-terminal domains. Our structures reveal that the N-terminal region of the C-terminal domain, a key component of the C-terminal transactivation domain, forms a long α helix that packs on the C-terminal end of the armadillo repeat domain, and thus forms part of the β-catenin superhelical core. The existence of this helix redefines our view of interactions of β-catenin with some of its critical partners, including ICAT and Chibby, which may form extensive interactions with this C-terminal domain α helix. Our crystallographic and NMR studies also suggest that the unstructured N-terminal and C-terminal tails interact with the ordered armadillo repeat domain in a dynamic and variable manner.

SUMMARY

Footnotes

ACCESSION NUMBERS

Coordinates and structure factors have been deposited in the Protein Data Bank (PDB; http://www.rcsb.org/pdb) under ID codes 2Z6G and 2Z6H.

SUPPLEMENTAL DATA

Supplemental Data include in vitro pull-down assays (Figure S1) and the structural superposition of the N-terminal helices of armadillo repeats (Figure S2) and are available at http://www.structure.org/cgi/content/full/16/3/478/DC1/.

Footnotes
Collaboration tool especially designed for Life Science professionals.Drag-and-drop any entity to your messages.