Inhibition of β-catenin/B cell lymphoma 9 −protein interaction using α-helix–mimicking sulfono-γ-AApeptide inhibitors

Peng Sanga,1, Min Zhangb,1, Yan Shia,1, Chunpu Lia,c,d, Sami Abdulkadira,QiLic,d,2, Haitao Jib,2, and Jianfeng Caia,2

aDepartment of Chemistry, University of South Florida, Tampa, FL 33620; bDrug Discovery Department, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL 33612; cDepartment of Medical Oncology, Shuguang Hospital, Shanghai University of Traditional Chinese Medicine, Shanghai 201203, China; and dAcademy of Integrative Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai 201203, China

Edited by Samuel H. Gellman, University of Wisconsin–Madison, Madison, WI, and approved April 22, 2019 (received for review November 16, 2018) The rational design of α-helix–mimicking peptidomimetics pro- compared with that of α-helix (5.4 Å), as well as the remarkable vides a streamlined approach to discover potent inhibitors for pro- stability of sulfono-γ-AApeptides, we envisioned that helical sul- tein−protein interactions (PPIs). However, designing cell-penetrating fono-γ-AApeptides could be adapted to develop a new class of helical long peptidomimetic scaffolds equipped with various functional mimetics that disrupt α-helix-mediated protein–protein interactions. groups necessary for interacting with large protein-binding interfaces With the aim of developing new PPIs for various biological remains challenging. This is particularly true for targeting β-catenin/ targets, we wondered whether B Cell Lymphoma 9 (BCL9) (25– BCL9 PPIs. Here we designed a series of unprecedented helical sul- 39), previously shown to engage its α-helical HD2 domain to fono-γ-AApeptides that mimic the binding mode of the α-helical HD2 interact with β-catenin (Fig. 2), could be used to develop pep- domain of B Cell Lymphoma 9 (BCL9). Our studies show that sul- tidomimetic helical foldamers based on sulfono-γ-AApeptides. fono-γ-AApeptides can structurally and functionally mimic the The Wnt/β-catenin signaling pathway plays important roles in α-helical domain of BCL9 and selectively disrupt β-catenin/BCL9 PPIs embryonic development and tissue homeostasis, as well as in with even higher potency. More intriguingly, these sulfono- several types of human cancer, such as colorectal cancer, breast γ-AApeptides can enter cancer cells, bind with β-catenin and disrupt –

cancer, melanoma, prostate cancer, and others (40 43). BIOCHEMISTRY β-catenin/BCL9 PPIs, and exhibit excellent cellular activity, which is As a central mediator of the signaling, β-catenin controls the ex- much more potent than the BCL9 peptide. Furthermore, our enzy- pression of several key that regulate the cell cycle and apo- matic stability studies demonstrate the remarkable stability of the ptosis. As the matter of fact, transcriptional activation of Wnt/ γ helical sulfono- -AApeptides, with no degradation in the presence β-catenin signaling pathway is dependent on formation of the of pronase for 24 h, augmenting their biological potential. This work β-catenin supercomplex involving BCL9 or BCL9-like (B9L), as well γ represents not only an example of helical sulfono- -AApeptides that as the T cell factor (Tcf)/lymphoidenhancer-bindingfactor(Lef) α – mimic -helix and disrupt protein protein interactions, but also an family of transcriptional factors (37). Specifically, BCL9 functions as a excellent example of potent, selective, and cell-permeable unnatural scaffolding structure of the Wnt enhanceosome and brings β-catenin β foldameric peptidomimetics that disrupt the -catenin/BCL9 PPI. The to TCF/LEF to transcribe specific Wnt target genes, leading to cell design of helical sulfono-γ-AApeptides may lead to a new strategy to – growth, proliferation, and differentiation (44). As such, molecules modulate a myriad of protein protein interactions. that disrupt BCL9/β-catenin protein–protein interaction could inhibit

α-helix mimetics | β-catenin | B-cell lymphoma 9 | protein–protein interactions | inhibitors Significance

The design of cell-penetrating long peptidomimetic scaffolds he development of peptidomimetic helical foldamers for for interacting with large protein binding interfaces is chal- applications in chemical biology and drug discovery has attrac- T lenging because they involve various necessary functional tedmuchinterestinthefieldofmedicinalchemistry.Helicalfol- groups. Here we report the first series of helical sulfono- damers are commonly featured with many attractive merits, such as γ-AApeptides that disrupt protein–protein interactions. Spe- enhanced resistance to proteolytic degradation and high sequence cifically, we discovered that these helical mimetics can struc- diversity (1–8), and have been explored extensively for inhibition of turally and functionally mimic the B Cell Lymphoma 9 (BCL9) protein–protein interactions (PPIs) (9–13). Nevertheless, the tar- helix and disrupt cancer-related β-catenin/BCL9 protein–pro- geting of intracellular still entails significant challenges (14– tein interaction in cells with excellent potency and specificity. 16), largely due to the limited availability of molecular frameworks Enzymatic stability studies demonstrate the remarkable sta- of peptidomimetic peptides. bility of the helical sulfono-γ-AApeptides, augmenting their As a new class of proteolytically stable peptidomimetics, γ biological potential. This strategy could be adapted to target a -AApeptides have emerged as effective peptidomimetics that myriad of other disease-related protein–protein interactions. play important roles in chemical biology and biomedical sciences – γ (17 19). Specifically, sulfono- -AApeptides have been shown to Author contributions: H.J. and J.C. designed research; P.S., M.Z., Y.S., and C.L. performed have excellent folding stability to adopt a series of helical research;P.S.,S.A.,Q.L.,H.J.,andJ.C.analyzeddata;andP.S.,H.J.,andJ.C.wrote structures with a well-defined hydrogen-bonding pattern (20–24). the paper. In sulfono-γ-AApeptides, one-half of the side chains are in- The authors declare no conflict of interest. troduced by sulfonyl chlorides, providing enormous chemical This article is a PNAS Direct Submission. diversity (Fig. 1A) (17–19). In particular, we recently reported Published under the PNAS license. the X-ray crystal structures of a series of homogeneous L-sul- 1P.S., M.Z., and Y.S. contributed equally to this work. γ fono- -AA foldamers (24), which form well-defined left-handed 2To whom correspondence may be addressed. Email: [email protected], Haitao.Ji@ 414 helices (Fig. 1 B and C). Intriguingly, the side chains of sul- moffitt.org, or [email protected]. fono-γ-AApeptides are aligned perfectly on the top of one another This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. with a pitch of 5.1 Å. Based on the precise 3D arrangement of 1073/pnas.1819663116/-/DCSupplemental. their side functional groups, their close similarity in helical pitch

www.pnas.org/cgi/doi/10.1073/pnas.1819663116 PNAS Latest Articles | 1of6 Downloaded by guest on September 24, 2021 A C multiple myeloma. It is envisioned that unnatural peptidomi- metic inhibitors of β-catenin/BCL9 PPI would be appealing be- cause they can mimic peptide helices while being highly resistant to proteolytic degradation. However, to the best of our knowl- edge, to date there has been no report on the development of potent unnatural peptidomimetic inhibitors to disrupt β-catenin/ BCL9 PPI. This is possibly due to the fact that known peptidic foldamers do not efficiently mimic the long α-helix, as well as to the limited availability of scaffolds and molecular frameworks. With the availability of helical sulfono-γ-AApeptide scaffold, B we questioned whether sulfono-γ-AApeptides can be designed to effectively disrupt the β-catenin/BCL9 PPI. If so, this would offer a new template for generating potent helical peptidomimetics that inhibit a variety of medicinally relevant PPIs. In view of the folding pattern (Fig. 2 C–F), we believe that our sulfono-γ-AApeptides can effectively mimic BCL9, because the side chains of sulfono- γ-AApeptides could be designed to mimic the critical residues of 5.1 Å the α-helix of BCL9 HD2 domain. Here we report the develop- ment of some unnatural peptidomimetics that are highly effective γ Fig. 1. (A) The chemical structure of sulfono- -AApeptides; a and b denote the for inhibiting β-catenin/BCL9 PPI. chiral side chain and the sulfonamido side chain from the building block, re- γ spectively. (B) The crystal structure of a sulfono- -AApeptide. (C)TopviewofB. Results and Discussion Design and Biological Activity of Sulfono-γ-AApeptides. Our design Wnt/β-catenin signaling transduction and thus could be developed as was straightforward. As shown in Fig. 2 C and D, the chiral side novel anticancer agents. chains 2a, 4a, 6a, 8a, and 10a are on the same face of the helical The crystal structure of the β-catenin/BCL9/TCF-4 ternary scaffold of sulfono-γ-AApeptides, and thus this face was chosen complex (37) revealed that the helical domain of BCL9 (351- to mimic those critical residues of the BCL9 helical domain. The 374) interacts with a binding groove in β-catenin (Fig. 2 A and B). position map of those residues (Fig. 2E)ontheα-helical scaffold The critical residues of BCL9, R359, L363, L366, I369, and L373, demonstrates that R359, L363, L366, and L373 are almost on the which are on the one face of the BCL9 helix, form both hydro- same line, while I369 is not. A close comparison of helical scaf- philic and hydrophobic contacts with the binding surface of folds of the sulfono-γ-AApeptide and the BCL9 peptide reveals β-catenin. Despite an unambiguous mechanism of action, the that 8b, rather than 8a, could closely mimic I369 (Fig. 2F). As 8b is design of potent intracellular inhibitors to block this PPI remains the sulfonyl side chain, we hypothesized that a methyl sulfonyl a challenge, attributed mainly to the interaction between BCL9 group would be sufficient, because the sulfonyl group sticks out and β-catenin, which is mediated by an ∼25-residue helical seg- more than chiral side chains on helical sulfono-γ-AApeptides. ment from BCL9. While several small-molecule inhibitors have We thus designed and synthesized a panel of sulfono- been designed to disrupt β-catenin/BCL9 PPIs, peptide inhibitors γ-AApeptides (Fig. 3). The first sulfono-γ-AApeptide sequence are scarce (25, 27–30, 34). Wang (33, 35) and Takada (32) that we designed, sequence 2 (Fig. 3), contains only 10 sulfono- reported triazole-stapled and olefin-stapled BCL9 L351−F374 γ-AA building blocks (comparable in length to a 20-mer pep- α-helical peptides, respectively. tide), as the crystal structure (Fig. 2 A and B) shows that the first While the cell permeability of triazole-stapled peptides was few residues of BCL9 do not interact with β-catenin directly. not discussed, the olefin-stapled BCL9 peptide was found to pass We next measured the binding affinity of sulfono-γ-AApeptide the cell membrane, disrupt the β-catenin/BCL9 PPI, and selec- 2 toward β-catenin using a fluorescence polarization assay. In our tively suppress the transcription of Wnt target genes (32). This assay, the binding affinity of BCL9 peptide 1 exhibited a Kd value olefin-stapled peptide also inhibited cancer cell growth, angio- of 0.97 μM (Table 1), consistent with previous reports (19, 29). genesis, and metastasis without any evident damage to normal Excitingly, the first sulfono-γ-AApeptide sequence 2 that we tissues in mouse xenograft models for colorectal carcinoma and designed, with a shorter length than the BCL9 peptide 1, had a

Fig. 2. (A and B) The α-helical HD2 domain of BCL9, which directly engages a surface groove of β-catenin, provided the template for structural stabilization by hydrocarbon stapling [ (PDB) ID code 2GL7]. (A) Cartoon representation of the residues of BCL9 (red) critical for binding to β-catenin, shown as sticks. (B) BCL9 shown as sticks, and β-catenin represented with the surface model. (C–F) Schematic representation of distribution of side chains from sulfono-γ-AApeptides. (C) Side view. (D) Top view, helical wheel. (E) Position map of critical residues of the BCL9 helix. (F) Position map of side chains of sulfono-γ-AApeptides designed to mimic residues in E.

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1819663116 Sang et al. Downloaded by guest on September 24, 2021 BIOCHEMISTRY

Fig. 3. Structures of sulfono-γ-AApeptides investigated for the disruption of β-catenin–BCL9 interaction. The critical side chains are shown in red.

Kd value of 0.43 μM, which is already twofold more affinitive to with the values for peptide 1 (Table 1). We hypothesized that β-catenin than BCL9 peptide 1. although the critical residue for interaction is the side chain 8b, The ability of the sulfono-γ-AApeptide 2 to act as a functional the less bulkier methyl group at 8b is expected to have less impact mimic of BCL9 peptide 1 to disrupt BCL9/β-catenin PPI was on steric hindrance than Ile side chain on the neighboring 8b, then examined by AlphaScreen assays (29, 30). As shown in which may ensure a closer interaction of the sequence with Table 1, the helical BCL9 peptide 1 could disrupt β-catenin/ β-catenin. Indeed, a bulkier group at 8b led to decreased binding 10 BCL9 PPI, with Ki and IC50 values of 1.13 μM and 1.28 μM, activity (sequence ). respectively, which again are in very good agreement with the It seems that strong hydrophobic interaction near the C- literature (25, 27–30). To our delight, the sulfono-γ-AApeptide 2 terminal region is critical, as a change in the groups at positions was found to disrupt β-catenin/BCL9, with Ki and IC50 values of 7b, 8a, and 9a to less hydrophobic groups or cationic groups led 0.64 μM and 0.74 μM, respectively, making it almost twofold to sequences 8, 9, and 11, which show inferior binding activity. more potent than BCL9 peptide 1. This initial success demon- The importance of these key side chains was further manifested strates the potential of sulfono-γ-AApeptides to mimic the long by sulfono-γ-AApeptide sequence 12 (Fig. 3 and Table 1), which α-helix, as well as their robust helical folding propensity (24). lacks several key side chains at positions 1b, 2a, 6a, and 10a, To investigate the importance of the side chains in the sul- resulting in complete loss of the ability to inhibit β-catenin/BCL9 fono-γ-AApeptide sequence, alanine scanning studies were car- PPIs. It should be noted that the binding affinity, Kd, from ried out based on sulfono-γ-AApeptide sequence 2 (Fig. 3, fluorescence anisotropy studies is highly consistent with the Ki sequences 3–8). We replaced each key residue with an Ala side and IC50 values obtained from AlphaScreen assays that measure chain at positions 1b, 2a, 6a, 8a, and 10a (marked with an asterisk). It appears that aminoethane (position 1b), Arg (position 2a), and β Leu (position 10a) have important roles in inhibiting β-catenin/ Table 1. Activity of peptides for the disruption of -catenin and BCL9 PPI, as an approximate 1.5-fold decrease in binding affinity BCL9 interaction 2 relative to sequence was caused by each Ala substitution (Table Peptide Kd, μMIC50, μM, ± SD Ki, μM, ± SD 1, sequences 3, 4,and7). The results for sulfono-γ-AApeptide ± ± sequence 3 indicate that positively charged side chains could affect 1 0.97 1.28 0.29 1.13 0.24 ± ± binding activity even if they were not involved in direct contact 2 0.43 0.74 0.15 0.64 0.11 ± ± with the β-catenin/BCL9-binding pocket. 3 0.86 1.04 0.21 0.92 0.17 ± ± It appears that Leu (position 6a) is the most critical group for 4 0.82 1.10 0.14 0.98 0.11 ± ± inhibiting β-catenin/BCL9 PPI (Fig. 3 and Table 1, sequence 5), 5 1.97 2.71 0.37 2.43 0.32 6 0.16 0.54 ± 0.13 0.46 ± 0.10 as the Ala substitution resulted in decreased binding affinity, ± ± with K , K , and IC values of 1.97 μM, 2.43 μM, and 2.71 μM, 7 0.70 1.00 0.10 0.89 0.08 d i 50 8 3.89 8.01 ± 0.82 7.20 ± 0.72 respectively. Interestingly, mutation of Ile (position 8a) to Ala in ± ± sequence 6 further improved the binding affinity and inhibitory 9 5.14 12.9 2.15 11.6 1.92 μ 10 2.51 4.07 ± 0.59 3.65 ± 0.51 activity, with Kd, Ki, and IC50 values of 0.16 M(approximately ± ± sixfold greater), 0.46 μM (approximately threefold greater), and 11 8.26 17.4 3.35 15.7 3.00 12 >10 >10 >10 0.54 μM (approximately threefold greater), respectively, compared

Sang et al. PNAS Latest Articles | 3of6 Downloaded by guest on September 24, 2021 a strong positive maximum around 210 nm, which is consistent with the previously reported CD spectra of helical sulfono- γ-AApeptides (24), suggesting that sequences 2–12 adopt a similar left-handed helical conformation. As anticipated, peptide 1, with a length of 23 residues, also adopted a helical confor- mation in solution.

Cell Permeability Test. The inhibition of intracellular PPIs remains challenging in chemical biology and drug discovery (14–16). An intracellular PPI inhibitor has to penetrate and cross the cell membrane, which may also be why developing peptidomimetic- based inhibitors for β-catenin/BCL9 PPI, which has extremely long sequences, is so challenging. To determine whether our newly developed sulfono-γ-AApeptide Fig. 4. (A) Proposed structure of sulfono-γ-AApeptide 6 with critical side inhibitors could permeate Wnt/β-catenin–dependent cancer cells, we chains 2a, 4a, 6a, 8b, and 10a shown in stick representation. (B) Overlay of 6 next examined the cellular uptake of fluorescein isothiocyanate with critical residues of the BCL9 helical peptide using PyMOL software. (C) (FITC)-labeled derivatives of our sulfono-γ-AApeptides 2, 3, 4, 6, Overlay of 6 with critical residues of BCL9 on the binding surface of β-catenin and 9 and the BCL9 peptide 1 in SW480 cells by confocal fluores- (PDB ID code 2GL7) using PyMOL software. cence microscopy (Fig. 6 and SI Appendix,Fig.S5). When SW480 cells were treated with FITC-labeled peptide 1, negligible green fluorescence was observed at 1 μMand10μM for 2 h, consistent the direct competition of all tested sequences for the interaction 1 of BCL9 peptide with β-catenin, suggesting that these sequences with the observation that peptide has poor cell permeability and bind to the same site on β-catenin. exhibits no cellular activity (30). However, it is surprising that when γ SW480 cells were treated with FITC-labeled sulfono-γ-AApeptides The excellent binding activity of sulfono- -AApeptides was 2 3 4 6 9 further rationalized by the modeling studies. The structure of , , , ,and at the same concentrations for 2 h (Fig. 6 and SI sequence 6 was built on the scaffold of the crystal structure (24) Appendix,Fig.S5), strong and evenly diffused intracellular green fluorescence was noted in cytoplasm, even at concentrations as low and then superimposed onto the BCL9 helical domain by over- μ γ laying the helical backbone orientations using PyMOL software as 1 M. These results suggest that sulfono- -AApeptides are highly (Fig. 4A) (45). As shown in Fig. 4 B and C, the side chains of cell-permeable, possibly due to the existence of multisulfonamide critical residues of the BCL9 peptide overlap very well with the groups on the molecular scaffold. As such, although the BCL9 1 side chains of 2a, 4a, 6a, 8b, and 10a of sulfono-γ-AApeptide 6. peptide is known for its cell impermeability as well as poor cellular β – As a result, the helical sulfono-γ-AApeptide 6 could tightly bind activity toward Wnt/ -catenin dependent cancer cells, we hypothe- γ to the groove of β-catenin through both hydrophilic and hydro- sized that our sulfono- -AApeptides could cross membranes and phobic interactions using these side chains. gain access to targets within the cytoplasm of living cells.

Circular Dichroism Measurements. We hypothesized that sulfono- MTS Cell Viability Assay. MTS tetrazolium cell viability assays were β γ-AApeptides should adopt well-defined helices in solution be- performed to assess the effect of -catenin/BCL9 inhibitors on cause they generally have better activity than the natural BCL9 the cell proliferation of colorectal cancer cells (SW480), which peptide 1, the natural binding partner of β-catenin. Thus, we exhibit hyperactive Wnt/β-catenin signaling (Table 2). Consistent performed circular dichroism (CD) spectroscopic studies to in- with previous reports, the regular BCL9 peptide 1 showed very > μ vestigate the helicity of regular peptide 1 and homogeneous poor activity, with an IC50 of 200 M. Intriguingly, the MTS assays sulfono-γ-AApeptides 2–12. The CD studies were performed in of sulfono-γ-AApeptides 2–7 showed that the sulfono-γ-AApeptides PBS buffer between 190 nm and 260 nm. As shown in Fig. 5, inhibited cancer cell proliferation in a dose-dependent manner. Se- 2–4 each sulfono-γ-AApeptide showed a marked Cotton effect, with quences exhibited excellent inhibitory activity, with IC50 values of ∼12 μM toward SW480 cells (Table 2). Note that these sequences were roughly threefold more selective toward SW480 cells than A549 cells that have normal β-catenin signaling, suggesting good specificity 1 γ 6 2 of the tested sulfono- -AApeptides. Interestingly, sequence ,which 15 3 4 5 10 6 1 2 )

-1 7 FITC DAPI Merged FITC DAPI Merged 5 8 1 μM

dmol 9 1 μM 2 0 10 11 10 μM -5 12 10 μM (deg cm -3 -10 3 4 x 10 FITC DAPI Merged FITC DAPI Merged [θ] -15 1 μM 1 μM -20 10 μM 10 μM 200 210 220 230 240 250 260 Wavelength (nm) Fig. 6. Confocal fluorescence microscopy images of SW480 cells treated Fig. 5. CD spectra of BCL9 peptide 1 and sulfono-γ-AApeptides 2–12 with 1 μM and 10 μM of the FITC-labeled peptide 1 and sulfono-γ-AApep- (100 μM) measured at room temperature in PBS buffer. tides 2–4 for 2 h. (Magnification, 630×.)

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1819663116 Sang et al. Downloaded by guest on September 24, 2021 Table 2. MTS assay to monitor the inhibitory activities of The Renilla luciferase reporter construct (pCMV-RL) served as sulfono-γ-AApeptides on the viability of cancer cells the internal control to normalize luciferase reporter signals and eliminate systematic errors such as cell viability, transfection MTS IC50, μM, ± SD effect, and others. The TOPFlash luciferase reporter assay was Hyperactive Normal performed on the most potent sulfono-γ-AApeptides 2–4. Con- β-catenin– signaling β-catenin– signaling sistent with cell proliferation studies, sulfono-γ-AApeptide inhib- Peptide SW480 cell A549 cell itors 2–4 suppressed the TOPFlash luciferase activity in SW480 in a dose-dependent manner (Fig. 7A), with estimated IC50 values of 1 >200 >200 μ ± ± 35.4, 26.5, and 20.6 M, respectively (SI Appendix, Fig. S6). At 212.52.86 39.9 4.75 50 μM, each compound inhibited >90% of luciferase activity. As 312.8± 2.90 43.7 ± 3.62 expected, the regular BCL9 peptide 1 did not show any activity. 416.4± 4.40 43.2 ± 5.60 Interestingly, in the follow-up FOPFlash luciferase reporter 5 132 ± 18.9 230 ± 27.5 assay, the inhibitory activity of sulfono-γ-AApeptides 2–4 drop- 665.4± 7.06 150 ± 19.2 ped significantly (Fig. 7B). At a concentration of 50 μM, luciferase 746.6± 5.57 120 ± 13.6 retained >85% of its activity. The Renilla internal control values were constant in all TOPFlash/FOPFlash assays (SI Appendix,TableS2). Taken together, these results suggest that sulfono-γ-AApeptide in- was most active in vitro, displayed relatively weak activity, possibly due hibitors 2–4 selectively inhibit Wnt/β-catenin signaling transactivation to some unknown side interactions. This initial encouraging result while not generally inhibiting other transcriptional pathways. prompted us to further study the effects on Wnt/β-catenin signaling by TOPFlash and FOPFlash luciferase reporter assays (25, 28, 30). Cellular Target Engagement. We conducted two experiments to examine whether these sulfono-γ-AApeptides can engage β-catenin TOPFlash/FOPFlash Luciferase Reporter Assays. Wnt-specific TOP- in the cellular context. Biotinylated 3 (3-Biotin) and biotinylated 4 (4- Flash/FOPFlash luciferase reporter assays were used to evaluate Biotin) (the best two compounds in luciferase reporter assays) were β – the effects of these compounds on -catenin dependent tran- synthesized and incubated with SW480 cell lysates. The proteins that scription. For the TOPFlash reporter construct, the firefly lu- bind with these two sequences were then pulled down by streptavidin- ciferase reporter was placed downstream of three wild-type conjugated beads and examined by Western blot analyses using Tcf-binding sites. For the FOPFlash reporter construct, the β-catenin–specific antibody. As shown in Fig. 8A,both3-Biotin and 4- BIOCHEMISTRY firefly luciferase reporter gene was placed downstream of three Biotin could effectively bind with β-catenin in SW480 cell lysates at a mutant Tcf-binding sites. The high expression of firefly luciferase concentration of 1 μM. in TOPFlash assays was controlled by tandem Tcf-binding sites. Coimmunoprecipitation (co-IP) experiments were also per- formed with Wnt/β-catenin hyperactive HCT116 cancer cells to evaluate the effects of 4 on disruption of the β-catenin/BCL9 PPI in 4 β 120 100 103 cells. As shown in Fig. 8B, inhibitor disrupted the -catenin/BCL9 A 120 100 96.2 103 102 100 100 77.6 PPI in a dose-dependent manner, while the input and immuno- 80 80 precipitation controls were constant in the different experiments. 60 60 40 40 Enzymatic Stability Study. Along with cell permeability, protease 20 20 7.75 stability of the sequences is critical for their biological activity. 0 0 Relative Activity (%) Activity Relative 02550100 (%) Activity Relative 012.52550 We further evaluated the proteolytic stability of helical 1 (µM) 120 2 (µM) 120 100 102 100 90.1 100 100 80 80 57.7 47.2 Input Control 3-Biotin 4-Biotin 60 60 A 40 40 9.12 7.09 20 20 0 0 Relative Activity (%) Activity Relative

012.52550 (%) Activity Relative 012.52550 3 (µM) 4 (µM)

120 100 B B 100 98.2 101 92.8 120 95.7 101 Control 10 30 μM 100 100 86.4 80 80 60 60 BCL9 40 40 20 20 0 IP β-catenin

Relative Activity (%) Activity Relative 0 (%) Activity Relative 012.52550 β-catenin 02550100 1 (µM) 2 (µM) BCL9 120 100 100 103 120 100 103 95.3 90.7 Input 100 100 72.5 80 80 10% cell lysate 60 60 β-Tubulin 40 40 20 20 0 0 Relative Activity (%) Activity Relative 012.52550 (%) Activity Relative 012.52550 Fig. 8. (A) SW480 cell lysate was incubated with 3-Biotin or 4-Biotin, followed 3 (µM) 4 (µM) by streptavidin pull-down experiments. The levels of β-catenin associated with 3-Biotin and 4-Biotin were analyzed by Western blot analysis. Input: 5% of cell Fig. 7. (A) Wnt-responsive TOPFlash luciferase reporter assay results of inhibi- lysate. (B) Co-IP experiments to evaluate the disruption of the β-catenin/BCL9 tors 1–4 in β-catenin–activated SW480 cells. (B) Wnt-responsive FOPFlash lucif- PPI by 4 in Wnt/β-catenin hyperactive cancer cells. IP, immunoprecipitation; erase reporter assay results of inhibitors 1–4 in β-catenin–activated SW480 cells. Input, 10% of the cell lysate. Each experiment was performed in duplicate.

Sang et al. PNAS Latest Articles | 5of6 Downloaded by guest on September 24, 2021 sulfono-γ-AApeptides 2–4 and BCL9 peptide 1 in pronase, a protein pull-down and co-IP experiments demonstrated that these mixture of broad scope endopeptidases and exopeptidases iso- sulfono-γ-AApeptides can bind with β-catenin and disrupt β-cat- lated from Streptomyces griseus (46). The assays were conducted enin/BCL9 PPIs in cells. This work also represents the successful by incubating 0.1 mg/mL of three lead compounds 2–4 and the application of unnatural peptidomimetics in disrupting β-catenin/ 1 regular peptide with 0.1 mg/mL of pronase in 100 mM am- BCL9 PPIs, which has long been considered a challenging target, monium bicarbonate buffer (pH 7.8) at 37 °C for 24 h. The providing a practical method for the development of novel folda- stability of the examined compounds was analyzed by HPLC-MS – 1 meric peptidomimetics that serve as proteolytically stable and cell- (SI Appendix, Figs. S12 S15). The control peptide was com- penetrating inhibitors for a myriad of PPIs. We believe this work pletely degraded by pronase, with no intact peptide remaining can expand the utility of sulfono-γ-AApeptides in the preparation of (SI Appendix, Fig. S12), which may explain why peptide 1 showed weak cell permeability and completely abandoned its cellular potent and cell-permeable peptidomimetic agents that will find activity. Strikingly, our linear sulfono-γ-AApeptides showed no many applications in chemical biology and biomedical sciences. – detectable degradation (SI Appendix, Figs. S13 S15), demon- Materials and Methods strating extraordinarily high stability against enzymatic degra- γ dation, augmenting their potential in therapeutic applications. Building blocks and sulfono- -AApeptides were synthesized following pre- viously reported methods. All other chemicals and solvents were purchased In summary, we report a series of unprecedented helical sulfono- 1 13 γ-AApeptides that mimic α-helix and disrupt PPIs. These unnatural from commercial sources and used as received. Hand C NMR spectra were recorded on a Varian INOVA 400 spectrometer. High-resolution mass spectra helical peptidomimetics are able to disrupt cancer-related β-catenin/ were obtained on an Agilent 6220 using electrospray ionization time-of-flight. BCL9 PPIs with excellent potency and specificity. The cell-based γ Synthesis, characterization, and biological experiments are described in detail studies indicated that sulfono- -AApeptides are cell-permeable and in SI Appendix. can effectively inhibit the growth of cancer cells with hyperactive β Wnt/ -catenin signaling. The TOPFlash/FOPFlash luciferase re- ACKNOWLEDGMENTS. This work was supported by National Science porter assays demonstrated that sulfono-γ-AApeptides can selec- Foundation CAREER Award 1351265 (to J.C.) and National Institutes of tively suppress transactivation of Wnt/β−catenin signaling. The Health Grant 1R01 GM112652-01A1 (to J.C.).

1. J. E. Jones et al., Length-dependent formation of transmembrane pores by 310-helical 25. M. Zhang, Z. Wang, Y. Zhang, W. Guo, H. Ji, Structure-based optimization of small- α-Aminoisobutyric acid foldamers. J. Am. Chem. Soc. 138, 688–695 (2016). molecule inhibitors for the β-Catenin/B-cell lymphoma 9 protein-protein interaction. 2. G. W. Collie et al., Shaping quaternary assemblies of water-soluble non-peptide he- J. Med. Chem. 61, 2989–3007 (2018). lical foldamers by sequence manipulation. Nat. Chem. 7, 871–878 (2015). 26. R. Fan et al., SOX7 suppresses Wnt signaling by disrupting β-Catenin/BCL9 interaction. 3. R. W. Cheloha, A. Maeda, T. Dean, T. J. Gardella, S. H. Gellman, Backbone modifica- DNA Cell Biol. 37, 126–132 (2018). tion of a polypeptide drug alters duration of action in vivo. Nat. Biotechnol. 32, 653– 27. K. B. Teuscher, M. Zhang, H. Ji, A versatile method to determine the cellular bio- 655 (2014). availability of small-molecule inhibitors. J. Med. Chem. 60, 157–169 (2017). 4. J. Buratto et al., Structure of a complex formed by a protein and a helical aromatic 28. J. A. Wisniewski, J. Yin, K. B. Teuscher, M. Zhang, H. Ji, Structure-based design of 1,4- oligoamide foldamer at 2.1 Å resolution. Angew. Chem. Int. Ed. Engl. 53, 883–887 dibenzoylpiperazines as β-Catenin/B-cell lymphoma 9 protein-protein interaction in- (2014). hibitors. ACS Med. Chem. Lett. 7, 508–513 (2016). 5. C. Mayer, M. M. Müller, S. H. Gellman, D. Hilvert, Building proficient enzymes with 29. M. Zhang, J. A. Wisniewski, H. Ji, AlphaScreen selectivity assay for β-catenin/B-cell foldamer prostheses. Angew. Chem. Int. Ed. Engl. 53, 6978–6981 (2014). lymphoma 9 inhibitors. Anal. Biochem. 469, 43–53 (2015). 6. P. S. P. Wang, J. B. Nguyen, A. Schepartz, Design and high-resolution structure of a β3- 30. L. R. Hoggard et al., Rational design of selective small-molecule inhibitors for β-cat- peptide bundle catalyst. J. Am. Chem. Soc. 136, 6810–6813 (2014). enin/B-cell lymphoma 9 protein-protein interactions. J. Am. Chem. Soc. 137, 12249– 7. M. Wolffs et al., Helical aromatic oligoamide foldamers as organizational scaffolds for 12260 (2015). photoinduced charge transfer. J. Am. Chem. Soc. 131, 4819–4829 (2009). 31. J.-J. Zhao et al., miR-30-5p functions as a tumor suppressor and novel therapeutic tool 8. Y. Hamuro, J. P. Schneider, W. F. DeGrado, De novo design of antibacterial β-Peptides. by targeting the oncogenic Wnt/β-catenin/BCL9 pathway. Cancer Res. 74, 1801–1813 J. Am. Chem. Soc. 121, 12200–12201 (1999). (2014). 9. A. Barnard et al., Selective and potent proteomimetic inhibitors of intracellular 32. K. Takada et al., Targeted disruption of the BCL9/β-catenin complex inhibits onco- protein-protein interactions. Angew. Chem. Int. Ed. Engl. 54, 2960–2965 (2015). genic Wnt signaling. Sci. Transl. Med. 4, 148ra117 (2012). 10. V. Azzarito, K. Long, N. S. Murphy, A. J. Wilson, Inhibition of α-helix-mediated 33. S. A. Kawamoto et al., Design of triazole-stapled BCL9 α-helical peptides to target the protein-protein interactions using designed molecules. Nat. Chem. 5, 161–173 (2013). β-catenin/B-cell CLL/lymphoma 9 (BCL9) protein-protein interaction. J. Med. Chem. 55, 11. M. Pelay-Gimeno, A. Glas, O. Koch, T. N. Grossmann, Structure-based design of in- 1137–1146 (2012). hibitors of protein-protein interactions: Mimicking peptide binding epitopes. Angew. 34. M. de la Roche et al., An intrinsically labile α-helix abutting the BCL9-binding site of Chem. Int. Ed. Engl. 54, 8896–8927 (2015). β-catenin is required for its inhibition by carnosic acid. Nat. Commun. 3, 680 (2012). 12. S. Fuchs et al., Proline primed helix length as a modulator of the nuclear receptor- 35. S. A. Kawamoto et al., Analysis of the interaction of BCL9 with β-catenin and devel- coactivator interaction. J. Am. Chem. Soc. 135, 4364–4371 (2013). opment of fluorescence polarization and surface plasmon resonance binding assays 13. J. T. Ernst, J. Becerril, H. S. Park, H. Yin, A. D. Hamilton, Design and application of an for this interaction. Biochemistry 48, 9534–9541 (2009). α-helix-mimetic scaffold based on an oligoamide-foldamer strategy: Antagonism of 36. M. de la Roche, J. Worm, M. Bienz, The function of BCL9 in Wnt/beta-catenin sig- the Bak BH3/Bcl-xL complex. Angew. Chem. Int. Ed. Engl. 42, 535–539 (2003). naling and colorectal cancer cells. BMC Cancer 8, 199 (2008). 14. A. J. Wilson Inhibition of protein-protein interactions using designed molecules. 37. J. Sampietro et al., Crystal structure of a β-catenin/BCL9/Tcf4 complex. Mol. Cell 24, Chem. Soc. Rev. 38, 3289–3300 (2009). 293–300 (2006). 15. T. L. Nero, C. J. Morton, J. K. Holien, J. Wielens, M. W. Parker, Oncogenic protein 38. S. Adachi et al., Role of a BCL9-related β-catenin-binding protein, B9L, in tumori- interfaces: Small molecules, big challenges. Nat. Rev. Cancer 14, 248–262 (2014). genesis induced by aberrant activation of Wnt signaling. Cancer Res. 64, 8496–8501 16. M. R. Arkin, Y. Tang, J. A. Wells, Small-molecule inhibitors of protein-protein inter- (2004). actions: Progressing toward the reality. Chem. Biol. 21, 1102–1114 (2014). 39. Y. Wang et al., The Wnt/β-catenin pathway is required for the development of leu- 17. Y. Shi et al., γ-AApeptides: Design, structure, and applications. Acc. Chem. Res. 49, kemia stem cells in AML. Science 327, 1650–1653 (2010). 428–441 (2016). 40. R. Nusse, H. Clevers, Wnt/β-catenin signaling, disease, and emerging therapeutic 18. P. Teng, Y. Shi, P. Sang, J. Cai, γ-AApeptides as a new class of peptidomimetics. modalities. Cell 169, 985–999 (2017). Chemistry 22, 5458–5466 (2016). 41. J. N. Anastas, R. T. Moon, WNT signalling pathways as therapeutic targets in cancer. 19. Y. Shi et al., One-bead-two-compound thioether Bridged macrocyclic γ-AApeptide Nat. Rev. Cancer 13, 11–26 (2013). screening library against EphA2. J. Med. Chem. 60, 9290–9298 (2017). 42. T. N. Grossmann et al., Inhibition of oncogenic Wnt signaling through direct targeting 20. H. Wu et al., The synthesis of head-to-tail cyclic sulfono-γ-AApeptides. Org. Biomol. of β-catenin. Proc. Natl. Acad. Sci. USA 109, 17942–17947 (2012). Chem. 13, 672–676 (2015). 43. L. Dietrich et al., Cell permeable stapled peptide inhibitor of Wnt signaling that 21. H. Wu, P. Teng, J. Cai, Rapid access to multiple classes of peptidomimetics from targets β-catenin protein-protein interactions. Cell Chem. Biol. 24, 958–968.e5 (2017). common γ-AApeptide building blocks. Eur. J. Org. Chem. 2014, 1760–1765 (2014). 44. L. M. van Tienen, J. Mieszczanek, M. Fiedler, T. J. Rutherford, M. Bienz, Constitutive 22. P. Teng et al., Right-handed helical foldamers consisting of De novo d-AApeptides. J. scaffolding of multiple Wnt enhanceosome components by Legless/BCL9. eLife 6, Am. Chem. Soc. 139, 7363–7369 (2017). e20882 (2017). 23. P. Teng et al., Hydrogen-bonding-driven 3D supramolecular assembly of peptidomi- 45. The PyMOL Molecular Graphics System, version 2.1.1 (Schrödinger LLC, 2015). metic zipper. J. Am. Chem. Soc. 140, 5661–5665 (2018). 46. D. F. Hook et al., The proteolytic stability of “designed” β-peptides containing 24. F. She et al., De novo left-handed synthetic peptidomimetic foldamers. Angew. Chem. α-peptide-bond mimics and of mixed α,β-peptides: Application to the construction of Int. Ed. Engl. 57, 9916–9920 (2018). MHC-binding peptides. Chem. Biodivers. 2, 591–632 (2005).

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1819663116 Sang et al. Downloaded by guest on September 24, 2021