Getting a Grip on the Undrugged: Targeting β‐Catenin with Fragment‐Based Methods

Abstract Aberrant WNT pathway activation, leading to nuclear accumulation of β‐catenin, is a key oncogenic driver event. Mutations in the tumor suppressor gene APC lead to impaired proteasomal degradation of β‐catenin and subsequent nuclear translocation. Restoring cellular degradation of β‐catenin represents a potential therapeutic strategy. Here, we report the fragment‐based discovery of a small molecule binder to β‐catenin, including the structural elucidation of the binding mode by X‐ray crystallography. The difficulty in drugging β‐catenin was confirmed as the primary screening campaigns identified only few and very weak hits. Iterative virtual and NMR screening techniques were required to discover a compound with sufficient potency to be able to obtain an X‐ray co‐crystal structure. The binding site is located between armadillo repeats two and three, adjacent to the BCL9 and TCF4 binding sites. Genetic studies show that it is unlikely to be useful for the development of protein–protein interaction inhibitors but structural information and established assays provide a solid basis for a prospective optimization towards β‐catenin proteolysis targeting chimeras (PROTACs) as alternative modality.


Protein purification
The construct for expression of β-catenin 141-305 (Amino acids 141-305 of uniprot ID: P35222) is based on the previously reported R4 construct 1

NMR spectroscopy
Ligand based 1D 1 H experiments were carried out on a Bruker Avance II 600 MHz spectrometer with a 5 mm cryo-QCI probe and z-gradients. An in house customized Tecan Freedom Evo liquid handler 3 was used to automatically prepare and transfer samples into 2.5 mm NMR tubes inserted in 96 in house modified Bruker MATCH systems to ensure identical incubation times. NMR samples were transported to the magnet with the Bruker Sample Rail system. Randomly combined mixtures of four fragments from 50 mM d6-DMSO stocks were tested at an individual concentration of 250 µM each in 25 mM sodium phosphate, 150 mM sodium chloride, 100 mM sodium nitrate, 0.5 mM TCEP pH 6.8 in D2O at 25 °C. A protein concentration of 10 µM βcatenin 141-305 was used. For STD-NMR experiments a Gaussian pulse train with a duration of 3 s was employed to accomplish selective irradiation and a 30 ms spin-lock pulse was used to suppress residual protein signals. The on and off resonance spectra were recorded in an interleaved fashion to minimize subtraction artifacts. The irradiation frequency for the on-and off resonance experiment were -100 Hz and -40 kHz, respectively. After acquisition, the difference spectrum was calculated by subtracting the two individual spectra. Compound binding was observed by signals arising in the difference spectrum. These were readily assigned by comparing the signals with the respective pre-recorded reference spectra of the individual compounds present in the mixture. The quality control of the fragment library included purity, identity, and buffer solubility measurements. To rule out false positives all library members were also tested for their tendency of self-association or micelle formation by STD experiments in the absence of protein.
For confirmation of hits obtained from STD-NMR by 2D 15 N TROSY NMR 4 a Bruker Avance III 600 MHz instrument with a 5 mm cryo-TCI probe and z-gradient was used. The experimental set up (automated sample preparation and transfer into the magnet by a Tecan Freedom Evo pipetting robot and a Bruker Sample Rail system, respectively) was identical. Each sample contained 25 uM 15 N labeled β-catenin 141-305 25 mM sodium phosphate, 150 mM sodium chloride, 100 mM sodium nitrate, 0.5mM TCEP pH 6.8 in H2O at 25 °C and 8% (v/v) D2O. The protein was incubated with 500 uM fragment in a 2.5 mm NMR tube at 25 °C and a d6DMSO concentration of 1%. Spectra 1D, GEM by STD and WaterLOGSY experiments were recorded at 298 K on an Avance III 700MHz spectrometer equipped with a cryogenically cooled 5mm TCI probe. Spectra were processed and analyzed with Topspin 3.5 (Bruker BioSpin). The 1D spectra were recorded with a double WATERGATE 6 suppression element with the number of scans set to 32. The WaterLOGSY pulse sequence was used as described before 7

Microscale Thermophoresis.
Fluorescence labeling of β-catenin 141-305 with the NT647 dye was achieved using the Monolith NT.115 Protein Labeling Kit RED-NHS according to the manufacturer's protocol (NanoTemper Technologies, Munich, Germany). The BCL9 peptide was used as positive control for assay development to achieve a reliably detectable change in the thermophoretic mobility (Fnorm).
Experimental conditions were optimized to increase signal to noise ratio and assay window (buffer conditions, protein concentration, temperature gradient by infrared laser (IR) on time, capillaries).
Final assay conditions were 50 nM NT647 labeled β-catenin 141-305 in PBS pH 6.8 supplemented with 100 mM NaNO3, 0.5 mM TCEP and 0.05% Tween-20 at 25° C. Fragment screening detected by MST was carried out as previously described 8,9 . In brief, 50 nM NT647-labeled β-catenin  was incubated with 500 M compound and 1% (v/v) d6-DMSO at 298K. To achieve fully automated sample preparation and data collection an in house modified Monolith NT.015 was combined with a Hamilton Microlab Star pipetting robot in collaboration with Nanotemper Technologies 8,9 . Manual inspection of all MST traces was used to reject non-standard traces stemming from protein aggregation, bleaching, or fluorescence quenching. Fnorm values (Fnorm=Fhot/Fcold) were calculated and the mean values of the duplicates were compared to the mean values of the DMSO negative control. Finally, fragment hits were identified by ∆∆Fnorm ≥ ∆Fnorm(2sd DMSO) with ∆∆Fnorm = │∆Fnorm(compound)-∆Fnorm(DMSO)│. Protein integrity over the entire FBS was monitored by the BCL9 peptide positive control. For Kd determination a dilution series was pipetted using the fully automated set up using 200 mM DMSO stocks to achieve 2 mM compound at 1% DMSO as the highest concentration.

Surface Plasmon Resonance (SPR)
A Biacore T200 instrument was used for SPR analysis,β-catenin 141-305 was pre-diluted to a concentration of 0.1 mg/mL with 10 mM sodium acetate buffer, pH 5.0 and immobilized at a density of 6000 -7000 response units on flow cell 2 of a Biacore CM5 chip. Carbonic anhydraseII served as a reference protein on flow cell 3 and flow cell 1 was kept blank and used as the reference surface. A mixture of 20 mM Tris, pH 7.5, 150 mM NaCl, 0.5 mM TCEP, 0.005% Tween20, 2 % dimethyl sulfoxide was used as assay buffer. After pre-equilibration of the chip with 10 blank injections, the Kd measurements were performed by injecting a concentration series of analyte (9 concentrations, 1:1 dilutions). BCL9 347-392 served as a positive control. Kd values were determined by global fitting of the steady-state response for each of the 9 concentrations to a 1:1 interaction model using the Biacore T200 evaluation software. Each compound was injected three times and average Kd values were calculated from the repeats. The reported data are the mean values of three independent experiments ± standard deviations.

Structural Analysis
Crystals of β-catenin 141-305 were obtained using the sitting drop vapor diffusion method. At 4° C 0.2 µL protein solution (13 mg/mL protein, 6.3 mM BI01450033) were mixed with 0.2µL reservoir solution containing 25% PEG 3350, 100 mM BIS-TRIS buffer pH 5.5 and 0.2 M Magnesium chloride. Brick-shaped crystals appeared after 24 hours and grew to a size of about 80 µM. Crystals were flash-frozen in liquid nitrogen in reservoir solution supplemented with 23% Ethylene glycol and 6% Di-Ethylene glycol. Diffraction data was collected in-house at a RIGAKU MICROMAX-003. Images were processed with autoPROC 10 and the structure was solved by molecular replacement using the previously solve structure (pdb ID: 3SLA) as model. Model building and refinement was performed with CCP4, COOT and autoBUSTER v.2.11.2.
(http://www.globalphasing.com) using standard protocols 11,12 . The final model was analyzed with MolProbity revealing residues in 99.4% in Ramachandran preferred regions. Statistics for the data collection and refinement can be found in Supplementary Table 2 and stereo images of the binding modes (wall-eye stereo) with the refined 2Fo-Fc electron can be found in Supplementary Fig. 7.

Luciferase reporter gene assay
For the dual luciferase assay, reverse transfection (0.4 x 10 5 cells, black 96 well plate) was performed using an active β-catenin control (S33A_S37A_T41A_S45A) and respective binding pocket mutants with co-transfection of a β-catenin/TCF-responsive TOPFlash luciferase reporter construct and constitutive active Renilla luciferase. 10 mM LiCl treatment was performed for 24h at 48 hours post transfection. Three days after transfection, the Dual-Glo® Luciferase Assay System (Promega, E2940) was applied following the manufacturer's instructions. Reporter activation obtained with transfection of 5 ng active β-catenin control was set to 100% as a reference.

Synthetic Chemistry
Compound 1 is commercially available at different vendors (e.g Aurora Fine Chemicals) and

General Methods
Each synthetic transformation was monitored and confirmed by HPLC-MS. Commercial starting materials were used without further purification. Solvents used for reactions were of commercial "dry"-or "extra-dry" or "analytical" grade. All other solvents used were reagent grade. Air-and moisture-sensitive reactions were performed under dry nitrogen or argon atmosphere with dried glassware. Commercial starting materials were used without further purification. All solvents used for reactions were "dry"-or "extra-dry" or "analytical" grade. Other solvents were reagent grade.

Analytical NMR Spectroscopy:
NMR experiments were recorded on a Bruker Avance HD 500 MHz spectrometer equipped with a TCI cryoprobe at 298 K. Samples were dissolved in 600 μL DMSO-d6 and TMS was added as an internal standard. 1D 1 H spectra were acquired with 30° excitation pulses and an interpulse delay of 4.2 sec with 64k data points and 20 ppm sweep width.

Chiral Separation:
Chiral separation was performed with a Sepiatec SFC system. The following conditions were used, 20 x 250 mm IC column (Daicel) at a flow of 70ml/min, 40°C column temperature, a backpressure of 200 bar, and an isocratic eluent ratio of 40% methanol and 60% CO2.

Synthesis of compounds 4 and 5
Enantiomers 4 and 5 were obtained from a chromatographic separation on a chiral phase of a racemic precursor and subsequent methylation of the single enantiomers as shown below:

Synthesis of 2-(4-Methoxy-benzylamino)-1-phenyl-ethanol
A solution of 10.00 g (42.0 mmol, 1.0 eq.) N-Boc-2-Hydroxy-2-Phenethylamine in 100 ml 1M HCl/EtOAc was stirred at room temperature overnight. The reaction mixture was filtered and the filter cake was dried under vacuum to afford the intermediate raw material that was dissolved in 100 ml dichloromethane. After the addition of 5.74 g (42.0 mmol, 1.0 eq.) para-Methoxy benzaldehyde the reaction mixture was stirred for 4 hours. Then 2.39 g (63.0 mmol, 1.5 eq) sodium borohydride were added and the mixture was kept stirring overnight. TLC showed complete conversion. To the reaction mixture was added 250 ml water and it was extracted three times with crude product was purified by normal phase chromatography to obtain 6.00 g (23.0 mmol, 55%) 2-(4-Methoxy-benzylamino)-1-phenyl-ethanol.
HPLC-MS analysis of the reaction mixture indicated full conversion. Water was added (50 ml) and the mixture was extracted with dichloromethane (3 x 50 mL). After the combined organic layers were dried over sodium sulfate the crude material was obtained after concentration under reduced pressure. The crude material (890 mg) was directly used in the subsequent step.

Synthesis of compounds 6 and 7
Enantiomers 6 and 7 were obtained from a chromatographic chiral separation on a chiral phase of the racemate accessible via the route shown below. (0.028 mmol, 0.1 eq) Pd(dppf)2Cl2•CH2Cl2, 15.9 mg (0.028 mmol, 0.1 eq.) Pd(dba)2 and 9.1 mg (0.14 mmol, 0.5 eq.) zinc powder were added and the mixture was irradiated in a microwave at 150 °C for 12 hours. HPLC-MS analysis showed complete conversion.
After combination of the organic phases they were dried over magnesium sulfate and concentrated under reduced pressure. The crude product was purified via RP-chromatography to yield 15.0 mg