Rational design of Harakiri (HRK)-derived constrained peptides as BCL-xL inhibitors

Using the HRK BH3 domain, sequence hybridization and in silico methods we show dibromomaleimide staple scanning can be used to inform the design of BCL-xL selective peptidomimetic ligands. These HRK-inspired reagents may serve as starting points for the discovery of therapeutics to target BCL-xL-overexpressed cancers.


Supplementary Data Figures
a Hot-spot residues are highlighted in bold; all peptides are N-terminal acetylated and C-terminal amidated; the residues participating in the formation of constraints are referred to as X 1 and X 2 , and highlighted in red; mS5 denotes 2-(4'-pentenyl)-glycine; b conditions: 20 mM Tris, 150 mM NaCl, pH 7.6, 50 nM tracer, 250 nM BCL-x L.     and BID/MCL-1 (PDB ID: 2KBW) (asterisks denote the missing residues in the BID/BCL-x L structure)

Experimental Methods
Solid phase peptide synthesis Rink amide resin, Oxyma, HOBt, DIC and Fmoc-L-amino acids were purchased from Fluorochem (UK). Dimethylformamide (DMF) was purchased from Sigma-Aldrich (UK). The designed peptides were synthesised using standard Fmoc-based solid-phase methods on a LibertyBlue microwave-assisted automated peptide synthesiser (CEM; Mathews, NC, USA). Microwave-assisted Fmoc deprotection was achieved by treatment with 20% piperidine in DMF; for peptides containing aspartic acid, 5% formic acid was added to minimise the formation of aspartimide. Amino acids were coupled with Oxyma/DIC under microwave assistance. Upon completion of automated peptide assembly, all peptides were acetylated except the precursors of fluorescent labelled tracers. To make fluorescent labelled tracers, N-terminally free peptides were coupled with Fmoc-Ahx-OH, followed by deprotection using 20% piperidine in DMF and final coupling with 5(6)-carboxyfluorescein (Fisher Scientific, UK) using HOBt/DIC at room temperature overnight. All peptides were cleaved by TFA/H 2 O/TIPS/DODT (92.5 : 2.5 : 2.5 : 2.5) for 3 hours at room temperature. Cleavage cocktail solutions were concentrated under nitrogen flow and crude peptides were precipitated with treatment of cold diethyl ether (diethyl ether: cleavage cocktail=10:1). Crude peptides were redissolved in acetonitrile/H2O (1:1), purified using a 1260 Infinity II preparative HPLC system (Agilent; Santa Clara, CA, USA) and then lyophilised to afford pure solid peptides for further study.

Incorporation of constraints
Maleimide Each purified linear peptide was dissolved in acetonitrile/phosphate buffer (ratio=3:7; peptide concentration: 4 mg/mL; phosphate buffer: 20 mM phosphate, 150 mM NaCl, pH = 7.8) and transferred into a 15 mL centrifuge tube. TCEP (2 equiv.) was added and the peptide was agitated for 30 mins to reduce any disulfide bonds. Thereafter, dibromomaleimide (solution in acetonitrile, 2 equiv.) was added and the peptide was agitated for 2 h. A colour change to yellow was observed during the reaction. The reaction was monitored by LC-MS. Upon completion of the reaction, peptide solutions were injected directly into preparative HPLC to isolate final maleimide-constrained products.

m-Xylene
After being dissolved and transferred as mentioned above, each peptide was treated with TCEP (solution in phosphate buffer, 2equiv.) for 30 mins to reduce disulfide bonds, followed by addition of 1,3-bis(bromomethyl)benzene (solution in DMF, 2 equiv.) and DIPEA (4 equiv.). The reacting solution was agitated for 2 h and monitored by LC-MS. Upon completion of the reaction, reaction mixtures were purified using preparative HPLC.
Hydrocarbon Fmoc-(S)-2-(4-pentenyl)glycine was synthesised as previously described and used as a building block for hydrocarbon stapled peptides. 1 After peptide assembly, the on-resin peptide was acetylated and transferred to a round bottom flask. Grubbs 1st generation catalyst (0.10 equiv.) was added, followed by addition of anhydrous dichloroethane (DCE, 5 mL). The reaction was stirred in the dark for 4 h under nitrogen flow and repeated by addition of fresh catalyst (0.1 equiv.). The resin bound peptide was washed with DCE (3 times), cleaved and then crude peptide was purified using prep-HPLC.

Virtual alanine scanning
The crystal structures of BAD/BCL-x L (PDB ID: 1G5J) and HRK/SPPV14 (PDB ID: 6XY4) were selected for virtual alanine scanning. The homology model of HRK/BCL-x L was generated and optimised in Rosetta (see below). The three structures were uploaded individually on the BAlaS server, 4 which is based on the Bristol University Docking Engine (BUDE). 5 Beclin-1/BCL-x L (PDB id: 2PL1) and MULE/MCL-1 (PDB id: 5C6H) were also uploaded for identification of hot-spot residues in the sequences. Contributions of >4.2 kJ mol-1 were considered as significant.

Generation of homology models:
The ligand peptides were extracted from the original structures in PyMol. The extracted ligands were aligned into the binding sites of the corresponding receptors in PyMol and the originally bound ligands were removed from the complexes. Thereafter, the generated structures were repacked in Rosetta to remove small clashes.

BAlaS modelling:
All structures which were used to carry out BAlaS modelling were uploaded on the Bude Alanine Scanning server (https://pragmaticproteindesign.bio.ed.ac.uk/balas/) and summarised in Table S3.

Table S3
A summary of structures that were used to carry out BAlaS modelling.

Structure
Ligand BCL-x L (extracted from 1G5J) Homology model

Circular dichroism
The peptides were dissolved in phosphate buffer (peptide concentration: 50 μM, 20 mM phosphate, 100 mM NaCl, pH 7.4) to a final concentration of 50 μM. The CD spectra were obtained on a Chirascan circular dichroism spectrometer (Applied Photophysics, UK), at 20 °C, using a 1 mm quartz cuvette. The parameters were set as follows: wavelength 195-260 nm, scan speed 5 nm min-1, step resolution 1nm. The spectra were averaged over triplicates after subtracting buffer baselines. The percentages of helicity were calculated based on the mean residue ellipticity at 222 nm using the following equation: 222 is the observed residue ellipticity at 222 nm calculated using: where n = number of residues; [θ] obs = measured value in mdeg; C = sample concentration (mol/L); l = path length of the cuvette in cm.
[θ] 0 is the mean residue ellipticity of a random coiled peptide and calculated as (2220-53T).
[θ] max is the theoretically maximum mean residue ellipticity of a n-mer helical peptide and equals to [θ] ∞ · (n-3)/n, where [θ] ∞ equals to (-44000 + 250T) (T is the temperature of peptide solutions in Celsius; 20 °C was used in the calculations in this study).

Protein expression
The pET28a HIS-SUMO BCL-x L (chimera with BCL-2, 1-209, missing 27-82) 1, 6 was overexpressed in the E. coli strain BL21 and the pET28a HIS-SUMO MCL-1 (172-327) was over-expressed in the E. coli strain Rosetta 2. Cells were resuspended in 25 mM Tris, 500 mM NaCl, pH 8.0 solution and lysed using sonication. The lysate was spun down and the supernatant was filtered before applied to a 5mL HisTrap column. The HisTrap was then washed with 10 CV of 25 mM Tris, 500 mM NaCl, 15 mM imidazole, pH 8.0 solution followed by elution with 25 mM Tris, 500 mM NaCl, 300 mM imidazole, pH 8.0 solution. The His-SUMO tag was cleaved in overnight dialysis (25 mM Tris, 250 mM NaCl, pH 8.0) at 4 °C with Smt3 protease, Ulp1. Uncleaved materials and residual HIS-SUMO tag were removed by reapplication to a HisTrap; the flow-through was collected and further purified using a Superdex 75 SEC system (GE healthcare; Chicago, IL, USA).

Fluorescence anisotropy
Fluorescence anisotropy assays were performed using 384-well plates (Greiner Bioone, UK). Each assay was performed in triplicates and fluorescence anisotropy was measured using an EnVisionTM 2103 MultiLabel plate reader (Perkin Elmer; Waltham, MA, USA). Excitation wavelength of 480 nm (30 nm bandwidth) and emission wavelength of 535 nm (30 nm bandwidth) were used for the FAM labelled peptides. All assays were performed in 20mM Tris, 150 mM NaCl, pH 7.6 solutions. Fluorescence anisotropy data were processed and analysed as previously described. 7 Briefly, the measured data of the P (perpendicular intensity) and S (parallel intensity) channels were subtracted by the corresponding control wells, and used to calculate the intensity and anisotropy using the following equations: Where I = total intensity; P = perpendicular intensity; S = parallel intensity; G = instrument factor which was set to 1.1 for all assays; r = anisotropy; L b = ligand bound fraction; λ = 1, [FL] = fluorescent ligand concentration; k = K d ; y = L b × Flu-tracer and x = added protein concentration. For analysis of direct titration FA assays, equations (3), (4) and (5) were used to calculate K d values. For competition FA assays, the average anisotropy and the average standard deviation of the values derived from equation (3) were calculated and fit to a sigmoidal logistic model (equation (6)) using Origin 2021.

Proteolysis
Each peptide was dissolved in Tris buffer (20 mM Tris, 150 mM NaCl, pH 7.5) to a final concentration of 150 μM. α-chymotrypsin or trypsin was prepared as a stock solution of 600 nM and added to the peptide solutions (n/n=1/25000). Then the mixtures were incubated at 37 °C, with aliquots removed after 0, 5, 15, 30, 60 and 90min. Test samples were quenched in the same volume of acetonitrile containing 2% TFA. All assays were repeated three times. The HPLC analysis was performed on an Agilent 1290 Infinity II HPLC analytical system (20 μL injection, 0.5 mL min-1 flow rate, 5-95% acetonitrile:water gradient, 10 min run time). The HPLC time points were converted into a percentage of the original substrate (S), and plotted as a natural logarithm (lnS) against time (t) in minutes. Half-lives (t 1/2 ) were calculated by dividing -ln (2)