Targeting the Tcf4 G13ANDE17 Binding Site To Selectively Disruptβ‑Catenin/T-Cell Factor Protein−Protein InteractionsZheng Huang, Min Zhang, Shawn D. Burton, Levon N. Katsakhyan, and Haitao Ji*

Department of Chemistry, Center for Cell and Genome Science, University of Utah, Salt Lake City, Utah 84112−0850, United States

*S Supporting Information

ABSTRACT: Selective disruption of protein−protein inter-actions by small molecules is important for probing thestructure and dynamic aspects of cellular network. It can alsoprovide new therapeutic targets. β-Catenin of the canonicalWnt signaling pathway uses the same positively chargedgroove to bind with T-cell factor (Tcf), cadherin, andadenomatous polysis coli (APC). The extravagant formationof β-catenin/Tcf interactions drives the initiation andprogression of many cancers and fibroses, while β-catenin/cadherin and β-catenin/APC interactions are essential forcell−cell adhesion and β-catenin degradation. In this study, aselective binding site that can differentiate β-catenin/Tcf, β-catenin/cadherin, and β-catenin/APC interactions was identified byalanine scanning and biochemical assays. A new peptidomimetic strategy that incorporates SiteMap and multiple-copysimultaneous search was used to design selective small-molecule inhibitors for β-catenin/Tcf interactions. A potent inhibitor wasdiscovered to bind with β-catenin and completely disrupt β-catenin/Tcf interactions. It also exhibits dual selectivity for β-catenin/Tcf over β-catenin/cadherin and β-catenin/APC interactions in both biochemical and cell-based assays. This studyprovides a proof of concept for designing selective inhibitors for β-catenin/Tcf interactions.

In the postgenomic era, a myriad of functional proteins havebeen characterized. Inside of the cell, there is a complex

network of protein−protein interactions (PPIs). The proteintends to use the same interface to interact with the variousother proteins in different organelles or cellular environments.Under pathological conditions, PPIs lose the compartmentalconstraint, and unexpected PPIs take place, often leading todisease. One challenge of biomedical research is to design asmall molecule to modulate a specific protein−protein interfacewhile leaving the other related PPIs unaffected. The selectivedisruption of β-catenin/Tcf PPIs, a key downstream effector ofthe canonical Wnt signaling pathway, represents such a case.The canonical Wnt signaling plays a fundamental role in

directing cell proliferation, differentiation, and survival.1 β-Catenin is the key mediator of this signaling pathway. Themutations in the regulatory genes of the canonical Wntsignaling pathway, such as those that encode adenomatouspolysis coli (APC), Axin, and/or the N-terminal phosphor-ylation sites of β-catenin, result in the stabilization of β-cateninand an increased level of nuclear β-catenin. Nuclear β-cateninforms a complex with T-cell factor (Tcf)/lymphoid enhancerfactor (Lef) to induce the overexpression of cell proliferation,migration, and survival genes, such as cyclin D1, c-myc, andsurvivin. Canonical Wnt signaling can also be activated throughepigenetic silencing of Wnt antagonist genes or the autocrineactivation of Wnt ligands, frizzled, and dishevelled (Dvl). Thehyperactivation of the canonical Wnt signaling pathway isstrongly implicated in the initiation and progression of manycancers and fibroses. In addition, canonical Wnt signaling is

aberrantly overactivated in cancer stem cells, which drive tumorgrowth, seed metastases, and cause cancer recurrence afterremission.2−4 Significant efforts have been made to identifyinhibitors for the canonical Wnt signaling pathway.5−7

Inhibitors of the upstream sites are less desirable becausethey can cause cross-regulatory effects on the β-catenin-independent Wnt pathways or perturb the function of β-catenin in cell−cell adhesion. Thus, it would be better toidentify inhibitors for the downstream sites of the canonicalWnt signaling pathway. The formation of the β-catenin/Tcfcomplex in the cell nucleus is the penultimate step of canonicalWnt signaling. Transcriptional overactivation of canonical Wnttarget genes is solely dependent on the formation of thiscomplex. Therefore, selective inhibition of β-catenin/Tcf PPIsrepresents an appealing therapeutic target. It is important thatthe inhibitor binds to coactivator β-catenin rather thantranscriptional factor Tcf, because Tcf is essential for theother signaling pathways,8 while β-catenin is specific forcanonical Wnt signaling.The crystal structures of β-catenin in complexes with Xenopus

Tcf3,9 human Tcf4,10−12 mouse Lef1,13 E-cadherin,14 andAPC15,16 have been solved. A comparison of these structuresreveals that β-catenin uses the same armadillo repeats to bindTcf (human Tcf4 residues 7−51), cadherin (human E-cadherin

Received: July 30, 2013Accepted: November 5, 2013Published: November 5, 2013

Articles

pubs.acs.org/acschemicalbiology

© 2013 American Chemical Society 193 dx.doi.org/10.1021/cb400795x | ACS Chem. Biol. 2014, 9, 193−201

residues 819−873), and APC (residues 1477−1519 of humanAPC 20-amino acid repeat 3, APC-R3). The binding modes ofTcf, cadherin, and APC in this region are identical. Biochemicalanalyses confirm that the binding modes of Tcf, cadherin, andAPC to β-catenin are mutually exclusive.17−20 Selectivitybecomes a major challenge when designing inhibitors for β-catenin/Tcf interactions. The inhibitor cannot block the β-catenin/cadherin interaction because this interaction is essentialfor the integrity of epithelial junctions. The blockage of thisprotein−protein complex alters cell−cell adhesion and impairsthe functions of normal stem cells. The inhibitor cannot blockthe β-catenin/APC interaction either because this interaction iscritical for β-catenin degradation. The blockage of this protein−protein complex leads to a larger pool of free β-catenin, andmore β-catenin enters the cell nuclei to overactivate thetranscription of Wnt target genes, thereby promoting newcancer formation. In addition, the dissociation constant (Kd) ofthe β-catenin/Tcf complex is 7−10 nM13,21 and much lowerthan those of the β-catenin/cadherin (Kd: 41−82 nM)20 and β-catenin/APC (Kd: 0.6−3.1 μM for APC-R3, the APC repeatwith the highest binding affinity) complexes,15,21 which furthercomplicates the design of selective β-catenin/Tcf inhibitors.The disruption of β-catenin/Tcf interactions has been thoughtundruggable for these reasons.5,12

Although extensive high-throughput screening (HTS) hasbeen performed to identify small-molecule inhibitors for β-catenin/Tcf PPIs, only four examples have been reported(Supplementary Figure 1).22−25 However, the binding modesof these HTS hits remain unknown, including whether theybind to β-catenin or Tcf, and where on the protein they bind.This has prohibited the optimization of these HTS hits. Theselectivity of these inhibitors for β-catenin/Tcf over β-catenin/E-cadherin and β-catenin/APC interactions was low.26 Hydro-carbon-stapled Axin α-helix, aStAx-35, was identified to bindwith the Axin-binding site of β-catenin and inhibit β-catenin/Tcf interactions (Supplementary Figure 1).27 Axin is thescaffolding protein for β-catenin phosphorylation. β-Catenin/Axin interactions maintain a low level of free β-catenin in thecytoplasm. The use of these hydrocarbon-stapled peptides hasthe potential to elevate the concentrations of β-catenin innormal cells. Recently, we reported a rational design of small-molecule inhibitor UU-T01 (Supplementary Figure 1) bybioisostere replacement.28 In this study, we demonstrated thattargeting the Tcf4 G13ANDE17 binding site can selectivelyinhibit β-catenin/Tcf PPIs. Small-molecule inhibitors with

excellent cell-based activities and high selectivity for β-catenin/Tcf over β-catenin/cadherin were identified.

■ RESULTS AND DISCUSSION

Identification of a Selective Binding Site for β-Catenin/Tcf over β-Catenin/E-Cadherin and β-Catenin/APC Interactions. A careful inspection of the crystalstructures of β-catenin reveals that the binding modes of thesequences flanking Tcf4 D16, E-cadherin D830, and APC-R3D1486 have subtle but very important differences. A kinkbetween armadillo repeats 9 and 10 of β-catenin forms abinding site (Figure 1a, hot spot 126). Tcf4 G13ANDE17 bindsto this site. There are four well-defined pockets for the Tcf4G13ANDE17 binding site: pocket A is lined with C429, N430,K435, H470, S473, and R474; pocket B is deep and lined withV511, C573, L539, I569, I507, and the side chain carbons ofK508; pocket C is a surface pocket defined by L519 and I579;and pocket D is another surface pocket that is lined with C466,E462, L506, A509, and the side chain carbons of K508. PocketsA and B are connected to pocket C. Pockets B and D areconnected by K508 and R469.Residue K435 in pocket A forms a salt bridge with D16 of

human Tcf4, D830 of human E-cadherin, and D1486 of humanAPC-R3. This interaction is important for anchoring Tcf/Lef,9,11,18,21,29−31 cadherin,9,31 and APC9,29,32 in place. How-ever, the binding modes adjacent to this salt bridge are differentamong the three protein−protein complexes. Tcf4 E17 forms asalt bridge with β-catenin K508. β-Catenin K508A or Tcf4E17A mutation significantly decreases the β-catenin/Tcfaffinity.18,21,28,31 Our previous study demonstrated that Tcf4D16E17 alone could effectively disrupt β-catenin/Tcf PPIs with aKi value of 396.59 μM.28 Small-molecule inhibitor UU-T01mimics Tcf4 D16E17 and exhibits a Ki value of 3.14 μM fordisrupting β-catenin/Tcf PPIs. The residues in E-cadherin andAPC-R3 that correspond to Tcf4 E17 are S831 and T1487,respectively. These two residues do not form a direct bondingwith β-catenin. The backbone amide of Tcf4 A14 forms an H-bond with the side chain carboxamide of N516. Deletionexperiments show that the sequences N-flanking the conservedAsp is critical for Tcf/Lef binding to β-catenin, including G13−T20 of human Lef129 and G8−A14 of Tcf411,30,31 (Figure 1b).The residues in E-cadherin and APC-R3 that correspond toTcf4 G13AN15 do not bond directly with β-catenin. The FPcompetitive inhibition assay was performed to evaluate the

Figure 1. Identification of a binding site in β-catenin for the selective disruption of β-catenin/Tcf interactions. (a) Pockets A−D of the Tcf GANDEbinding site. Tcf4 G13ANDE17 is shown yellow as the reference. (b) Sequences of human Lef1 (Q9UJU2), Tcf4 (Q9NQB0), E-cadherin (P12830),and APC-R3 (P25054). The key binding segments of Lef1, Tcf4, E-cadherin, and APC-R3 are aligned and underscored. The N-terminal flankingamino acids are shown. θ and Φ are the hydrophobic and aromatic residues, respectively. The sequences that bind to the Tcf4 G13ANDE17 bindingsite are shown in a rectangular box colored red. (c) Results of the FP competitive inhibition assay for the unlabeled human Tcf4, E-cadherin, andAPC-R3 peptides. Each set of data is expressed as mean ± standard deviation (n = 3).

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb400795x | ACS Chem. Biol. 2014, 9, 193−201194

importance of the corresponding sequences of Tcf, cadherin,and APC, as shown in Figure 1c and Supplementary Figure 2.Tcf4 peptide (residues 7−51) has an IC50 value of 1.09 nM.28

Under the same assay condition, Tcf4 peptide (residues 7−51,D16A/E17A)28 or a truncated Tcf4 peptide (residues 18−51)cannot disrupt β-catenin/Tcf4 interactions even at a concen-tration of >200 μM. In contrast, the simultaneous mutation offour residues, P826YDS829 of human E-cadherin (residues 819−873) into alanine, only leads to a 10-fold increase of the IC50

value. The mutation of four residues, D1484ADT1487 of humanAPC-R3 (residues 1477−1519) into alanine, causes a 30-foldincrease of the IC50 value. The FP saturation binding assay wasalso performed for E-cadherin (819−873) and E-cadherinmutant (819−873, P826YDS829/A826AAA829). The Kd values ofE-cadherin and the E-cadherin P826YDS829/A826AAA829 muta-tion are 80.40 ± 1.1 and 832.1 ± 14.4 nM, respectively(Supplementary Figure 3). As a comparison, the Kd value ofhuman Tcf4 D16A/E17A is >100 μM, while the Kd value ofnative human Tcf4 is 4.71 nM.28 On the basis of these results,we proposed that targeting the Tcf GANDE binding site bymimicking the binding mode of Tcf4 G13ANDE17 would lead tothe generation of selective small-molecule inhibitors specific forβ-catenin/Tcf interactions.Peptidomimetic Strategy To Design Small-Molecule

Inhibitors for β-Catenin/Tcf interactions. Pockets A, B, andC of the Tcf GANDE binding site were characterized bySiteMap33 and multiple-copy simultaneous search (MCSS).34

SiteMap uses a GRID-based method to define the binding site,calculate hydrophobicity, and determine the orientations of H-bond donors and acceptors. The SiteMap analysis indicates thatthe hydrophobic region is mainly located in pocket B, as shownin Figure 2a. The molecular interaction fields (MIFs) of H-bond donors are mainly from the side chains of K508, R469,K435, S473, R474, R515, R612, and the backbone amide ofN430. The MIFs of H-bond acceptors were determined by theside chains of N516, E571, H578, S473, and the backbonecarbonyl of K508 (Supplementary Figure 4). The druggabilityscore, Dscore,33 was determined. Dscore is a linear function ofthree pocket properties: size, enclosure, and hydrophobicity. Alarge validation study indicated that the average of Dscore forundruggable, difficult, and druggable pockets were 0.827, 0.995,and 1.091, respectively.33 The Dscore of the Tcf4 G13ANDE17

binding site of β-catenin was 0.926, indicating that this is adifficult target for designing potent inhibitors.

Ten aromatic, aliphatic, polar, and charged functional groupscommonly found in the bioactive small molecules were used inMCSS analysis (Supplementary Table 1). The minima with thesecond most favorable interaction energy with benzene and thethird most favorable interaction energy with indole are locatedin pocket B (Figure 2b), where it forms hydrophobicinteractions with V511 and I569. The minima with the thirdfavorable interaction energy with benzene and the second mostfavorable interaction energy with indole are located in thearginine channel formed by R474 and R515, where thefavorable cation−π interactions were observed. In pocket C,benzene (minimum No. 5) and indole (minimum No. 4) formcation−π interactions with R612 and R582 and are stabilized byH578 and I569. In pocket D, benzene (minimum No. 6) andindole (minimum No. 5) are stabilized by R469, A509, C466,and L506. The MCSS results for the aliphatic apolar fragments,cyclohexane, propane, and isobutane, are described inSupplementary Figure 5. The region displaying the mostfavorable interaction energies with cyclohexane and isobutane islocated in pocket B lined by V511, I569, G512, and the sidechain carbons of K508. The polar functional groups, dimethylether, methanol, and N-methylacetamide, were used to explorethe H-bond capacity in these pockets (Supplementary Figure6). The polar functional groups, methanol and N-methylacetamide, contain H-bond donors. The lowest energy minimaof methanol (minimum No. 9) and N-methylacetamide (minimNo. 4) as H-bond donors form H-bonds with side chaincarbonyl oxygen of N516, mimicking the H-bond betweenbackbone amide bond of Tcf4 A14 and side chain carbonyloxygen of β-catenin N516.On the basis of the results of the above SiteMap and MCSS

analyses, the peptidomimetic strategy was employed to designnew β-catenin/Tcf inhibitors. The G13ANDE17 sequence ofhuman Tcf4 was used as the starting template. The previousstudies indicated that D16 and E17 of human Tcf4 were criticalfor Tcf to bind with β-catenin.28 These two residues were keptin new inhibitors. The MCSS analysis shows that indole can belocated in the R474 and R515 arginine channel. It was mergedto the crystallographic binding conformation of Tcf4G13ANDE17, which leads to the deletion of residues Gly-Ala.The electron-rich indole ring was used to form cation−πinteractions with the arginine channel. The commerciallyavailable hydrophobic D- and L-amino acids identified fromSciFinder search were used to replace Asn to explorehydrophobic pocket B. AutoDock 4.235 was employed to

Figure 2. Analysis of the Tcf4 G13ANDE17 binding site. (a) Results of hydrophobic SiteMap analysis. The threshold of the SiteMap contour was setto −0.5 kcal/mol. (b) Representative MCSS-minimized positions of hydrophobic aromatic fragments, benzene (yellow), and indole (green). β-Catenin (PDB id: 2GL7) is shown as a space-filled model. Pockets A, B, C, and D are indicated.

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb400795x | ACS Chem. Biol. 2014, 9, 193−201195

predict the binding mode of new compounds. Compound 2 inFigure 3a was selected as the first designed compound forsynthesis. The predicted binding mode of 2 with β-catenin isshown in Figure 3b. The AutoDock model indicated that themethoxytyrosine moiety of 2 points toward hydrophobicpocket B. The indole ring of 2 can form cation−π interactionswith R515 (parallel geometry) and R474 (T-shaped geometry).The NH of the indole ring can form an H-bond with the sidechain carbonyl of N516. The indole ring of 2 can also interactwith surface hydrophobic pocket C. The predicted bindingmode of the designed compound matches the proposed keybinding elements. Compound 1, AcGANDEOMe, wassynthesized as a reference compound. Compound 3 wasdesigned to evaluate the effect of the ester and amide groups oninhibitory activity. Three new fragments, 4, 6, and 7a, and aknown fragment, 5,28 were used to evaluate the contribution ofeach substructure of 2 to inhibitory activity. Compounds 1−7awere synthesized in the first round. The synthetic routes for 1−7a are shown in Supplementary Schemes 1−4.The FP competitive inhibition assay results of 1−7a are

shown in Figure 4a and Supplementary Figure 7a. Compound 2exhibits a Ki value of 5.74 ± 0.76 μM and can completelydisrupt the large and tight β-catenin/Tcf PPIs. Compared to 1(Ki = 290.4 ± 1.3 μM), compound 2 is 50-fold more potent,

indicating the success of the design. The inhibitory activity ofamide derivative 3 is slightly less potent than 2. The same trendwas observed for the Asp−Glu dipeptide with the ester (4) andamide (5) termini. Fragments 6 and 7 also exhibited someinhibitory activities but were much less potent than 2. The FPassay also demonstrated that 2 was more potent than knowninhibitors KF115−584, CGP049090, PNU74654, iCRT3,iCRT5, and iCRT14 and comparable to PKF115−310 andUU-T01.

Exploration of the Arginine Channel and Pocket B.Further structural modification was the exploration of thestructure−activity relationship (SAR) for the indole moiety of2. Compounds 8−11 were designed. The synthetic routes for8−11 are shown in Supplementary Schemes 1 and 5. The FPcompetitive inhibition assay results are shown in Figure 4b andSupplementary Figure 7b. The Ki of 2 is lower than those of 8and 9, indicating chlorine is a better substituent. Theimportance of the indole moiety of 2 was further evaluatedby the synthesis of 3,4-disubstituted benzacetamide derivatives10 and 11. The AutoDock study shows that 10 and 11 do nothave an NH to form an H-bond with the side chain carbonyl ofN516 and have weaker cation−π interactions with R474 andR515 due to fewer π-electrons. In accordance with ourexpectation, the Ki values of 10 and 11 were over 60-fold

Figure 3. Structure-based design of new inhibitors based on Tcf4 G13ANDE17. (a) Structural evolution from 1 to 2. (b) AutoDock predictedconformation of 2 (gray). β-Catenin is shown as a space-filled model. Human Tcf4 G13ANDE17 (green) is shown as the reference.

Figure 4. FP competitive inhibition assay results of new inhibitors. Each set of data is expressed as mean ± standard deviation (n = 3). (a) FP Kivalues of 1−7a and eight known β-catenin/Tcf inhibitors. (b) FP Ki values of 8−14. (c) FP Ki values of 15−20.

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb400795x | ACS Chem. Biol. 2014, 9, 193−201196

higher than that of 2, indicating the importance of the H-bondand cation−π interactions for inhibitor binding. To investigatethe effect of the distance between the amide bond and theindole ring and the orientation of the indole ring on inhibitorypotency, compounds 12−14 were designed and synthesized(Supplementary Scheme 6). In good agreement with thedocking results, substitutions at position 2 of the indole ring (8and 12) yield a better inhibitor than those at position 3 (13 and14). Compared to 12, a relatively flexible 2-substituted sidechain, as in 8, can significantly increase the inhibitory activitiesbecause the indole can stay comfortably in the arginine channel,and the NH of the indole ring can form a desired H-bond withN516.The deep and narrow hydrophobic pocket B was explored by

substituted tyrosine derivatives and the other unnatural aminoacids. Compounds 15−20 were designed and synthesized(Supplementary Schemes 1 and 7). The FP assay results of 15−20 are shown in Figure 4c and Supplementary Figure 7c.Isopropyl derivative 15 has the lowest inhibitory activities inthis series. The isopropyl group is too large for pocket B. Thephenyl group of 16 and the cyclohexyl group of 17 only stay atthe entrance of pocket B, exhibiting Ki values lower than that of2. More hydrophobic derivatives, 18−20, have better inhibitoryactivities. Among this series, 19 exhibits the highest inhibitoryactivity for β-catenin/Tcf PPIs (Ki = 1.36 ± 0.12 μM). It is>210-fold more potent than 1 and represents one of the mostpotent inhibitors discovered to date.AlphaScreen and Isothermal Titration Calorimetry

(ITC) Studies. The AlphaScreen assay was used to evaluate 2and 18−20, as shown in Figure 5a and Supplementary Figure 8.AlphaScreen is a bead-based technique. It is similar to the time-resolved fluorescence resonance energy transfer (TR-FRET)technique and can provide orthogonal results with respect tothe FP assay. In the AlphaScreen assay, the streptavidin-coated

donor beads and nickel-chelate acceptor beads were broughttogether through C-terminally biotinylated human Tcf4(residue 7−51) and N-terminally His6-tagged human β-catenin(residues 138−686) interaction. The Ki of 19 in theAlphaScreen assay is 1.32 ± 0.56 μM. This compound ismore potent than known inhibitors CGP049090 and PKF118−310 in the parallel assays. The ITC studies indicated that 19binds to wild type β-catenin but not Tcf (Figure 5b andSupplementary Figure 9). The Kd value of 19 for β-catenin is0.418 μM, indicating that 19 binds to β-catenin and disrupts β-catenin/Tcf interactions. Figure 5c shows an AutoDockpredicted binding mode of 19 with β-catenin.

Inhibitor Selectivity Studies. The FP selectivity assayestablished in our previous studies26 was used to quantifyinhibitor selectivity between β-catenin/Tcf, β-catenin/E-cadherin, and β-catenin/APC interactions. The results areshown in Table 1 and Supplementary Figure 10. Theselectivities of 19 for β-catenin/Tcf over β-catenin/E-cadherininteractions and β-catenin/APC interactions are more than175- and 63-fold, respectively. To our best knowledge,compound 19 is the most selective inhibitor reported to date.

Site-Directed Mutagenesis Studies. The binding modeof 19 with β-catenin was evaluated by site-directed mutagenesis.As shown in Supplementary Figure 11, the V511S, V511S/I569S, R474A, R515A, and R474A/R515A mutants do notaffect β-catenin/Tcf binding. The FP competitive inhibitionassay was performed to derive the Ki values of 19 for wild typeβ-catenin/Tcf and mutant β-catenin/Tcf interactions. The Kivalue of 19 to wild type β-catenin/Tcf PPIs is 1.36 ± 0.12 μM.The Ki values of 19 to β-catenin V511S/Tcf4 and β-cateninV511S/I569S double mutant/Tcf4 interactions are 5.68 ± 0.78μM and 36.72 ± 0.77 μM, respectively (Table 2 andSupplementary Figure 12). It suggests that a change of pocketB from hydrophobic to hydrophilic greatly reduces the potency

Figure 5. Further validation of new potent β-catenin/Tcf inhibitors. (a) AlphaScreen Ki values of 2 and 18−20 and two known inhibitors. Each setof data is expressed as mean ± standard deviation (n = 3). (b) ITC study result of 19 (UU-T02) with β-catenin (T = 303.15 K). (c) AutoDockmodel of 19 with β-catenin.

Table 1. FP Selectivity Assay to Determine Inhibitor Selectivity of Potent Small-Molecule β-Catenin/Tcf Inhibitors

Ki ± SD (μM) selectivity

compounds β-catenin/Tcf4 β-catenin/E-cadherin β-catenin/APC-R3 Tcf/cadherin Tcf/APC

PKF115−58426 17.78 ± 2.13 13.04 ± 0.31 53.85 ± 0.08 0.7 3.0CGP04909026 35.56 ± 4.40 14.39 ± 0.32 62.49 ± 0.05 0.4 1.8PKF118−31026 5.82 ± 0.24 13.30 ± 0.11 171.3 ± 0.1 2.3 29.4UU-T0126 3.14 ± 0.48 103.6 ± 0.4 176.0 ± 0.1 33.0 56.02 5.74 ± 0.76 136.0 ± 4.2 77.21 ± 1.75 23.7 13.518 4.30 ± 0.59 158.0 ± 2.3 115.0 ± 2.9 36.8 26.719 1.36 ± 0.12 238.1 ± 2.3 86.63 ± 0.64 175.1 63.7

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb400795x | ACS Chem. Biol. 2014, 9, 193−201197

of 19, demonstrating the importance of hydrophobicity ofpocket B for inhibitor potency. R474 and R515 of the argininechannel were mutated to alanine to examine the role of thepositively charged guanidino side chain in ligand binding. TheKi values of 19 to β-catenin R474A/Tcf4 and β-catenin R515A/Tcf interactions are 11.05 ± 1.24 and 4.69 ± 0.75 μM,respectively, suggesting the guanidino group of R474 is moreimportant than that of R515 for binding with 19. The deletionof two gaunidino groups of R474 and R5151 leads to a drasticreduction of the inhibitory activity of 19 (Ki = 52.14 ± 1.24 μMfor β-catenin R474A/R515A double mutant/Tcf4 interactions),suggesting that the importance of the guanidino groups in thearginine channel for inhibitor binding. The same trend of wasobserved for compounds 2 and 18 (Table 2 and SupplementaryFigure 12).Cell-Based Studies. The Wnt-responsive luciferase re-

porter assay was performed using two cell lines, Wnt-activatedhuman embryonic kidney cell (HEK) 293 and colorectal cancercell SW480. HEK293 cells were co-transfected with β-catenin

plasmid and TOPFlash (luciferase reporter gene with wild typeTcf4 binding sites) or FOPFlash (luciferase reporter gene withmutant Tcf4 binding sites). SW480 cells were only transfectedwith TOPFlash or FOPFlash because this cell type has highlyelevated β-catenin. Compound 21 (Figure 6a), an ethyl esterderivative of 19, was synthesized (Supplementary Scheme 1).Compound 21 exhibited the IC50 values of 28.73 ± 3.09 and37.57 ± 3.23 μM in luciferase reporter assays for SW480 andWnt-activated HEK293 cells, respectively, while not reducingthe FOPFlash luciferase activities (Figure 6b and Supplemen-tary Figure 13), which implicates the specificity of thisinhibition.AXIN2 and LGR5 are the specific targets for the canonical

Wnt signaling pathway. Cyclin D1 and c-myc are two importanttarget genes of canonical Wnt signaling that promotetumorigenesis. As shown in Figure 6c, compound 21 down-regualted the transcription of AXIN2, LGR5, cyclin D1, and c-myc in dose-dependent manners in SW480 cells. The proteinexpression levels of cyclin D1, c-myc, and β-catenin in SW480cells were also examined by the Western blot analysis (Figure6d). The protein expression levels of cyclin D1 and c-myc weresignificantly reduced after the treatment of 21. In contrast, theprotein expression level of β-catenin was not affected, indicatingthat 21 does not inhibit the upstream sites of canonical Wntsignaling.Co-immunoprecipitation experiments were performed to

evaluate inhibitor selectivity in colorectal cancer cell lineHCT116. As shown in Figure 6e, compound 21 inhibited β-catenin/Tcf PPIs in a dose-dependent manner but had no affecton β-catenin/cadherin and β-catenin/APC PPIs at the highestexamined concentration. The MTs cell viability assay was

Table 2. Site-Directed Mutagenesis Studies and the FP AssayTo Evaluate the Binding Mode of 2, 18, and 19

Ki (μM)

β-catenin 2 18 19

wild type 5.74 ± 0.76 4.30 ± 0.59 1.32 ± 0.56V511S 32.47 ± 2.08 25.98 ± 0.80 5.68 ± 0.78V511S/I569S 135.2 ± 2.1 83.69 ± 1.64 36.72 ± 0.77R474A 44.00 ± 1.80 23.28 ± 1.71 11.05 ± 1.24R515A 35.94 ± 1.04 37.30 ± 0.90 4.69 ± 0.75R474A/R515A 240.6 ± 2.2 145.8 ± 1.8 52.14 ± 1.75

Figure 6. Cell-based studies of 21 (UU-T03). Each set of data is expressed as mean ± standard deviation (n = 3). (a) Chemical structure of 21. (b)Wnt-responsive luciferase reporter assay to monitor the inhibitory activities of 19, 21, and quercetin. (c) Quantitative real time PCR study todetermine the changes of mRNA expression of AXIN2, LGR5, cyclin D1, and c-myc in response to different concentrations of 21. (d) Western blotanalysis to monitor the changes of protein expression of c-myc, cyclin D1, and β-catenin in response to different concentrations of 21. β-Tublin wasused as an internal reference. (e) Co-immunoprecipitation experiments to evaluate inhibitor selectivity of 21 for β-catenin/Tcf over β-catenin/cadherin and β-catenin/APC PPIs. IP, immunoprecipitation; IB, immunoblotting; input, 10% amount of cell lysate. (f) MTs assay to monitor theeffects of 19, 21, and quercetin on the growth of colorectal cancer cells. n.d., not determined.

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb400795x | ACS Chem. Biol. 2014, 9, 193−201198

performed to assess the effect of 21 on the growth of colorectalcancer cell lines SW480, HCT116, and HT29. SW480 andHT29 harbor the deletion of APC, while HCT116 harbors amutation on β-catenin that blocks β-catenin phosphorylationand ubiquitination. These cancer cell lines were chosen on thebasis of their known dependence on β-catenin for growth andsurvival. Compound 21 was found to inhibit the growth ofcolorectal cancer cells with the IC50 values of 10.77 ± 1.18,11.01 ± 1.15, and 28.83 ± 1.26 μM for colorectal cancer cellsSW480, HT29, and HCT116, respectively (Figure 6f). Thiscompound exhibited a much higher IC50 value of 84.02 ± 3.62μM for normal cell line HEK293, again demonstrating theselectivity of this inhibitor. A known inhibitor of the canonicalWnt signaling pathway, quercetin,36 was used as reference inthe cell-based assays.Discussion. The design of a selective inhibitor that only

disrupts one protein−protein interface, while leaving all of theother related protein−protein interfaces unaffected, is challeng-ing. This challenge is even higher when the desired PPIs to bedisrupted have a much higher binding affinity than the PPIs tobe maintained. The Kd value of the native β-catenin/Tcfcomplex is 7−10 nM, while those of the β-catenin/E-cadherinand β-catenin/APC-R3 complexs are 41−82 nM and 0.6−3.1μM, respectively. In this study, we analyzed the crystallographicbinding modes of β-catenin with Tcf, cadherin, and APC andidentified a binding site that is critically important for theselective disruption of β-catenin/Tcf interactions throughalanine scanning and the FP assays.The peptidomimetic strategy faces challenges when design-

ing small-molecule PPI inhibitors due to the large protein−protein surface and the noncontiguous distribution of hotspots.37 It was stated that the peptides derived from shortcontiguous sequences at the protein−protein interface are poorchemical starting points for disrupting PPIs. The currentpeptidomimetic strategies, however, strive to mimic protein hotspot residues28,38 or protein secondary structures,39 includingα-helix, β-sheet, and β-turn. We hypothesized that apeptidomimetic strategy could be used to generate new potentinhibitors through the exploration of new pockets/new bindingelements that are adjacent to a hot spot but not used by PPIs.To achieve this goal, we have established a new peptidomimeticstrategy incorporating SiteMap and MCSS to solve thisproblem in this study. Both the SiteMap and MCSS analysesindicate the existence of a new hydrophobic pocket B (Figure2a). The hydrophobicity of this pocket is due to the side chainsof I569, L539, and V511. The SiteMap and MCSS analyses alsoidentified two additional new binding elements: (1) R474 andR515 form an arginine channel that can stabilize cation-πinteractions; (2) a surface hydrophobic pocket C that is definedby L519 and I579. Based on Tcf4 G13ANDE17, a series ofpeptidomimetic compounds were designed to take advantage ofthe hydrophobic nature of pockets B and C and the positivecharge of the arginine channel. The SAR analysis of 19derivatives (2−20) and the site-directed mutagenesis resultswere consistent with the proposed binding mode of theinhibitors. The cell-based studies demonstrated that 19 had alow cell-based activity, which was caused by the two carboxylicanions. The ester derivative, 21, can effectively pass the cellmembrane, inhibit the transactivation of canonical Wntsignaling, and down-regulate the expression of target genesthat are specific for the canonical Wnt signaling pathway.Compound 21 has no effect on the level of soluble β-cateninprotein but significantly decreases the β-catenin/Tcf association

in colorectal cancer cell SW480. Compound 21 also exhibitshigh cell-based selectivity for β-catenin/Tcf over β-catenin/cadherin and β-catenin/APC PPIs even though Tcf, cadherin,and APC bind to the same groove of β-catenin. Compound 21also exhibits good cell-based inhibitory activities against thegrowth of colorectal cancer cells and selectivity for cancer cellsover normal cells. This study provides a proof of concept fordesigning selective small-molecule inhibitor for β-catenin/Tcfinteractions, a highly challenging PPI target.

Conclusion. The discovery of small-molecule inhibitors thatselectively disrupt one protein−protein interface not onlyprovides a class of new therapeutic targets but also greatlyfacilitates our understanding of the structure and dynamicaspects of cellular networks. The extravagant formation of β-catenin/Tcf interactions in the downstream sites of canonicalWnt signaling drives the initiation and progression of manycancers and fibroses. β-Catenin/cadherin and β-catenin/APCinteractions, on the other hand, are essential for cell−celladhesion and maintaining a low level of free β-catenin innormal cells. β-Catenin uses the same large interface interactingwith Tcf, cadherin, and APC. The β-catenin/Tcf interaction ismuch stronger than the β-catenin/cadherin and β-catenin/APCinteractions. In this study, we used alanine scanning and the FPassays to identify a selective binding site that is criticallyimportant for β-catenin/Tcf interactions but much lessimportant for β-catenin/cadherin and β-catenin/APC inter-actions. A new peptidomimetic strategy incorporating SiteMapand MCSS was used to explore new pockets/new bindingelements adjacent to the hot spot. Hydrophobic pocket B andthe positively charged arginine channel composed of β-cateninR474 and R515 are employed to design new potent andselective inhibitors. After three rounds of structural optimiza-tion, compound 19 was identified. Site-directed mutagenesisand structure−activity relationship studies were performed toevaluate the binding mode of 19. This compound has a Kdvalue of 0.418 μM for binding to β-catenin and a Ki value of1.36 μM to completely disrupt β-catenin/Tcf interactions. Italso exhibits dual selectivity for β-catenin/Tcf over β-catenin/cadherin and β-catenin/APC interactions. Compound 21, anethyl ester derivative of 19, can effectively pass the cellmembrane and inhibit canonical Wnt signaling and the growthof colorectal cancer cells. This compound provides an excellentstarting point to generating more potent and selectiveinhibitors specific for β-catenin/Tcf interactions. Furthermore,the selectivity profile of 19 and 21 for β-catenin/Tcf over β-catenin/cadherin and β-catenin/APC interactions can be moreoptimized. Since pockets B and C are only important for β-catenin/Tcf interactions, we envision that further modificationof 19 by better occupying pockets B and C will lead to newinhibitors with better selectivity. It will also lead to higherinhibitor potency. The current study also includes the swappingof these two carboxylic groups of 19 with more lipophiliccarboxylic group bioisosteres discovered in our previousstuidy28 to improve inhibitor potency and cell permeability.

■ METHODSFragment Docking Using MCSS. The functional groups chosen

for the MCSS analysis were benzene, indole, cyclohexane, propane,isobutane, dimethyl ether, methanol, trans-N-methyl acetamide,acetate anion, and methylammonium cation. The binding site areain the MCSS simulation was defined to include all the residues of theTcf4 G13ANDE17 binding site. 2500−5000 replicas of a givenfunctional group were randomly distributed inside the binding site

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb400795x | ACS Chem. Biol. 2014, 9, 193−201199

(the distance threshold between fragment and protein was set to 0.9Å) and then simultaneously and independently energy minimized. Adistance-dependent dielectric model was used to approximate thesolvent. Default iteration profile and energy thresholds were used.Pairs of molecules were considered to be identical if the rmsd betweenthem was less than 0.2 Å. The minima with the binding energy lowerthan 0 kcal/mol were retrieved in the analysis. All of the calculationswere performed using the CHARMM force field and Momany andRone partial charge.40 At the end of the MCSS run, energy minimawithin 2.0 Å rmsd of each other were considered as one cluster, andthe copy in each cluster with the lowest energy was selected as clusterrepresentative.Ligand Docking Using AutoDock 4.2.35 In AutoDock studies,

only the polar hydrogen atoms remained on the protein structure, andKollman united atom charges were assigned. The 3D structures of theligands were built, and the partial atomic charges were calculated usingthe Gasteiger−Marsili method. The rotatable bonds in the ligandswere defined using AutoTors, which also united the nonpolarhydrogens and partial atomic charges to the bonded carbon atoms.The grid maps were calculated using AutoGrid. The AutoDock areawas defined to include all the residues of the Tcf4 G13ANDE17 bindingsite, and the grid spacing was set to 0.375 Å. Docking was performedusing the Lamarckian genetic algorithm, and the pseudo-Solis andWets method was applied for the local search. Each dockingexperiment was performed 100 times, yielding 100 dockedconformations. Other settings were the standard default parameters.All of the ligands followed the same docking protocol. The results ofthe docking experiments were evaluated by auxiliary clustering analysisand a visual inspection to match the proposed binding elements.FP Competitive Inhibition Assay.41,26 Experiments were

performed in 96-well Microfluor 2 black plates on a Synergy 2 platereader (Biotek). The polarization was measured at RT with anexcitation wavelength at 485 nm and an emission wavelength at 535nm. The FP experiments were performed in an assay buffer of 137 mMNaCl, 2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4, 100 μg/mL ofbovine gamma globulin, and 0.01% Triton-X 100. The final reactionvolume was set to 100 μL. For the β-catenin/Tcf assay, 10 nMconcentration of β-catenin (residues 142−686) was incubated with 2.5nM C-terminally fluorescein-labeled human Tcf4 (residues 7−51) for30 min at 4 °C, and then different concentrations of the testedpeptides and compounds in assay buffer were added. The negativecontrol (equivalent to 0% inhibition) refers to 2.5 nM Tcf4fluorescence tracer and 10 nM β-catenin in assay buffer without thetested compound. The positive control (equivalent to 100%inhibition) refers to only 2.5 nM Tcf4 fluorescence tracer in assaybuffer. For the β-catenin/cadherin assay, 150 nM human β-catenin(residues 138−686) was incubated with 5 nM C-terminallyfluorescent-labeled human E-cadherin (residues 819−873) in assaybuffer for 30 min at 4 °C. The negative control refers to 5 nM E-cadherin fluorescence tracer and 150 nM β-catenin in assay buffer withno inhibitor presenting. The positive control refers to 5 nM E-cadherinfluorescence tracer in assay buffer. For the β-catenin/APC-R3 assay,2000 nM human β-catenin (residues 138−686) was incubated with 5nM of C-terminally fluorescent-labeled human APC-R3 (residues1477−1519) in assay buffer for 30 min at 4 °C. The negative controlrefers to 5 nM APC-R3 fluorescence tracer and 2000 nM β-catenin inassay buffer without the tested compound. The positive control refersto 5 nM APC-R3 fluorescence tracer in assay buffer. Each assay platewas covered black and gently mixed on an orbital shaker at 4 °C for2.5 h to reach equilibrium before the polarization values were read.The background of the tested inhibitors was corrected by subtracting

the raw intensity values of the sample background well (allcomponents except probe) from the raw intensity values of thecorresponding test wells (all components). The IC50 values weredetermined by GraphPad Prism 5.0. The Ki values were derived fromthe IC50 values.

26 All of the experiments were performed in triplicateand carried out in the presence of 1% DMSO for small-moleculeinhibitors. The results were expressed as mean ± standard deviation.The Tcf/cahderin selectivity ratios were calculated on the basis of therespective Ki value of β-catenin/E-cadherin interactions over that of β-catenin/Tcf4 interactions. The Tcf/APC selectivity ratios werecalculated on the basis of the respective Ki value of β-catenin/APC-R3 interactions over that of β-catenin/Tcf4 interactions.

Site-Directed Mutagenesis Studies. β-Catenin mutants V511S,V511S/I569S, R474A, R515A, and R474A/R515A were generatedusing the overlapping PCR technique. The template for themutagenesis reactions was the wild type full-length β-catenin inpET-28a. KOD hot start DNA polymerase (Novagen) was usedthrough all the experiments. Mutants were confirmed by directsequencing (Core facility, University of Utah). The templates used forβ-catenin double mutation were V511S for V511S/I569S and R474Afor R474A/R515A, respectively. The double mutants were againconfirmed by direct sequencing. Following the confirmation of thesequence, β-catenin mutants were cloned into a pET-28 vector andtransformed into E. coli BL21 DE3. Primers for mutants are given inTable 3.

■ ASSOCIATED CONTENT*S Supporting InformationSupplementary figures, supplementary table 1, supplementaryschemes for synthesizing 1−21, and supplementary methodsfor the AlphaScreen, site-directed mutagenesis, ITC, and cell-based studies, and the synthetic procedures. This material isavailable free of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: markji@chem.utah.edu.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was in part supported by the Pulmonary FibrosisFoundation I. M. Rosenzweig Young Investigator Award(Award number, 235170). We thank D. P. Goldenberg at theUniversity of Utah for the use of his isothermal titrationcalorimetry instrument, and the Center for High-PerformanceComputing at the University of Utah for computer time.

■ REFERENCES(1) Clevers, H., and Nusse, R. (2012) Wnt/β-catenin signaling anddisease. Cell 149, 1192−1205.(2) Malanchi, I., Peinado, H., Kassen, D., Hussenet, T., Metzger, D.,Chambon, P., Huber, M., Hohl, D., Cano, A., Birchmeier, W., andHuelsken, J. (2008) Cutaneous cancer stem cell maintenance isdependent on β-catenin signalling. Nature 452, 650−653.(3) Barker, N., Ridgway, R. A., van Es, J. H., van de Wetering, M.,Begthel, H., van den Born, M., Danenberg, E., Clarke, A. R., Sansom,

Table 3

mutants primer 1 primer 2

I569S gtacaaccttcaacgctttcttccatg catggaagaaagcgttgaaggttgtacV511S caagatttcgaatcaatccagaagtagcctttatcagaggc gcctctgataaaggctacttctggattgattcgaaatcttgR474A ctgcttcttggtgtgcgctggtcagatg catctgaccagcgcacaccaagaagcagR515A aagggcaagatttgcaatcaatccaac aagggcaagatttgcaatcaatccaac

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb400795x | ACS Chem. Biol. 2014, 9, 193−201200

O. J., and Clevers, H. (2009) Crypt stem cells as the cells-of-origin ofintestinal cancer. Nature 457, 608−611.(4) Yeung, J., Esposito, M. T., Gandillet, A., Zeisig, B. B., Griessinger,E., Bonnet, D., and So, C. W. E. (2010) β-Catenin mediates theestablishment and drug resistance of MLL leukemic stem cells. CancerCell 18, 606−618.(5) Barker, N., and Clevers, H. (2006) Mining the Wnt pathway forcancer therapeutics. Nat. Rev. Drug Discovery 5, 997−1014.(6) Polakis, P. (2012) Drugging Wnt signalling in cancer. EMBO J.31, 2737−2746.(7) Anastas, J. N., and Moon, R. T. (2013) WNT signalling pathwaysas therapeutic targets in cancer. Nat. Rev. Cancer. 13, 11−26.(8) Weber, B. N., Chi, A. W.-S., Chavez, A., Yashiro−Ohtani, Y.,Yang, Q., Shestova, O., and Bhandoola, A. (2011) A critical role forTCF-1 in T-lineage specification and differentiation. Nature 476, 63−68.(9) Graham, T. A., Weaver, C., Mao, F., Kimelman, D., and Xu, W.(2000) Crystal structure of a β-catenin/Tcf complex. Cell 103, 885−896.(10) Poy, F., Lepourcelet, M., Shivdasani, R. A., and Eck, M. J. (2001)Structure of a human Tcf4-β-catenin complex. Nat. Struct. Biol. 8,1053−1057.(11) Graham, T. A., Ferkey, D. M., Mao, F., Kimelman, D., and Xu,W. (2001) Tcf4 can specifically recognize β-catenin using alternativeconformations. Nat. Struct. Biol. 8, 1048−1052.(12) Sampietro, J., Dahlberg, C. L., Cho, U. S., Hinds, T. R.,Kimelman, D., and Xu, W. (2006) Crystal structure of a β-catenin/BCL9/Tcf4 complex. Mol. Cell 24, 293−300.(13) Sun, J., and Weis, W. I. (2011) Biochemical and structuralcharacterization of β-catenin interactions with nonphosphorylated andCK2-phosphorylated Lef-1. J. Mol. Biol. 405, 519−530.(14) Huber, A. H., and Weis, W. I. (2001) The structure of the β-catenin/E-cadherin complex and the molecular basis of diverse ligandrecognition by β-catenin. Cell 105, 391−402.(15) Ha, N.-C., Tonozuka, T., Stamos, J. L., Choi, H.-J., and Weis, W.I. (2004) Mechanism of phosphorylation-dependent binding of APCto β-catenin and its role in β-catenin degradation. Mol. Cell 15, 511−521.(16) Xing, Y., Clements, W. K., Le Trong, I., Hinds, T. R., Stenkamp,R., Kimelman, D., and Xu, W. (2004) Crystal structure of a β-catenin/APC complex reveals a critical role for APC phosphorylation in APCfunction. Mol. Cell 15, 523−533.(17) Hulsken, J., Birchmeier, W., and Behrens, J. (1994) E-cadherinand APC compete for the interaction with β-catenin and thecytoskeleton. J. Cell Biol. 127, 2061−2069.(18) Omer, C. A., Miller, P. J., Diehl, R. E., and Kral, A. M. (1999)Identification of Tcf4 residues involved in high-affinity β-cateninbinding. Biochem. Biophys. Res. Commun. 256, 584−590.(19) Orsulic, S., Huber, O., Aberle, H., Arnold, S., and Kemler, R.(1999) E-cadherin binding prevents β-catenin nuclear localization andβ-catenin/LEF-1-mediated transactivation. J. Cell Sci. 112, 1237−1245.(20) Choi, H.-J., Huber, A. H., and Weis, W. I. (2006)Thermodynamics of β-catenin-ligand interactions: the roles of theN- and C-terminal tails in modulating binding affinity. J. Biol. Chem.281, 1027−1038.(21) Knapp, S., Zamai, M., Volpi, D., Nardese, V., Avanzi, N., Breton,J., Plyte, S., Flocco, M., Marconi, M., Isacchi, A., and Caiolfa, V. R.(2001) Thermodynamics of the high-affinity interaction of TCF4 withβ-catenin. J. Mol. Biol. 306, 1179−1189.(22) Lepourcelet, M., Chen, Y.-N. P., France, D. S., Wang, H., Crews,P., Petersen, F., Bruseo, C., Wood, A. W., and Shivdasani, R. A. (2004)Small−molecule antagonists of the oncogenic Tcf/β-catenin proteincomplex. Cancer Cell 5, 91−102.(23) Trosset, J.-Y., Dalvit, C., Knapp, S., Fasolini, M., Veronesi, M.,Mantegani, S., Gianellini, L. M., Catana, C., Sundstrom, M., Stouten, P.F. W., and Moll, J. K. (2006) Inhibition of protein−proteininteractions: the discovery of druglike β-catenin inhibitors bycombining virtual and biophysical screening. Proteins 64, 60−67.

(24) Gonsalves, F. C., Klein, K., Carson, B. B., Katz, S., Ekas, L. A.,Evans, S., Nagourney, R., Cardozo, T., Brown, A. M., and DasGupta, R.(2011) An RNAi-based chemical genetic screen identifies three small-molecule inhibitors of the Wnt/wingless signaling pathway. Proc. Natl.Acad. Sci. U.S.A. 108, 5954−5963.(25) Tian, W., Han, X., Yan, M., Xu, Y., Duggineni, S., Lin, N., Luo,G., Li, Y. M., Han, X., Huang, Z., and An, J. (2012) Structure-baseddiscovery of a novel inhibitor targeting the β-catenin/Tcf4 interaction.Biochemistry 51, 724−731.(26) Zhang, M., Catrow, J. L., and Ji, H. (2013) High-throughputselectivity assays for small-molecule inhibitors of β-catenin/T-cellfactor protein−protein interactions. ACS Med. Chem. Lett. 4, 306−311.(27) Grossmann, T. N., Yeh, J.-H. T., Bowman, B. R., Chu, Q.,Moellering, R. E., and Verdine, G. L. (2012) Inhibition of oncogenicWnt signaling through direct targeting of β-catenin. Proc. Natl. Acad.Sci. U.S.A. 109, 17942−17947.(28) Yu, B., Huang, Z., Zhang, M., Dillard, D. R., and Ji, H. (2013)Rational design of small-molecule inhibitors for β-catenin/T-cell factorprotein−protein interactions by bioisostere replacement. ACS Chem.Biol. 8, 524−529.(29) von Kries, J. P., Winbeck, G., Asbrand, C., Schwarz-Romond, T.,Sochnikova, N., Dell’Oro, A., Behrens, J., and Birchmeier, W. (2000)Hot spots in β-catenin for interactions with LEF-1, conductin andAPC. Nat. Struct. Biol. 7, 800−807.(30) Fasolini, M., Wu, X., Flocco, M., Trosset, J.-Y., Oppermann, U.,and Knapp, S. (2003) Hot spots in Tcf4 for the interaction with β-catenin. J. Biol. Chem. 278, 21092−21098.(31) Gail, R., Frank, R., and Wittinghofer, A. (2005) Systematicpeptide array-based delineation of the differential β-catenin interactionwith Tcf4, E-cadherin, and adenomatous polyposis coli. J. Biol. Chem.280, 7107−7117.(32) Liu, J., Xing, Y., Hinds, T. R., Zheng, J., and Xu, W. (2006) Thethird 20 amino acid repeat is the tightest binding site of APC for β-catenin. J. Mol. Biol. 360, 133−144.(33) Halgren, T. A. (2009) Identifying and characterizing bindingsites and assessing druggability. J. Chem. Inf. Model. 49, 377−389.(34) Zoete, V., Meuwly, M., and Karplus, M. (2005) Study of theinsulin dimerization: binding free energy calculations and per-residuefree energy decomposition. Proteins 61, 79−93.(35) Morris, G. M., Huey, R., Lindstrom, W., Sanner, M. F., Belew, R.K., Goodsell, D. S., and Olson, A. J. (2009) AutoDock4 andAutoDockTools4: Automated docking with selective receptorflexibility. J. Comput. Chem. 30, 2785−2791.(36) Park, C. H., Chang, J. Y., Hahm, E. R., Park, S., Kim, H.-K., andYang, C. H. (2005) Quercetin, a potent inhibitor against β-catenin/Tcfsignaling in SW480 colon cancer cells. Biochem. Biophys. Res. Commun.328, 227−234.(37) Wells, J. A., and McClendon, C. L. (2007) Reaching for high-hanging fruit in drug discovery at protein−protein interfaces. Nature450, 1001−1009.(38) Buckley, D. L., Van Molle, I., Gareiss, P. C., Tae, H. S., Michel,J., Noblin, D. J., Jorgensen, W. L., Ciulli, A., and Crews, C. M. (2012)Targeting the von Hippel-Lindau E3 ubiquitin ligase using smallmolecules to disrupt the VHL/HIF-1α interaction. J. Am. Chem. Soc.134, 4465−4468.(39) Whitby, L. R., and Boger, D. L. (2012) Comprehensivepeptidomimetic libraries targeting protein−protein interactions. Acc.Chem. Res. 45, 1698−1709.(40) Momany, F. A., and Rone, R. (1992) Validation of the generalpurpose QUANTA® 3.2/CHARMm® force field. J. Comput. Chem.13, 888−900.(41) Zhang, M., Huang, Z., Yu, B., and Ji, H. (2012) Newhomogeneous high-throughput assays for inhibitors of β-catenin/Tcfprotein−protein interactions. Anal. Biochem. 424, 57−63.

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb400795x | ACS Chem. Biol. 2014, 9, 193−201201