A cell-active cyclic peptide targeting the Nrf2/Keap1 protein–protein interaction

The disruption of the protein–protein interaction (PPI) between Nrf2 and Keap1 is an attractive strategy to counteract the oxidative stress that characterises a variety of severe diseases. Peptides represent a complementary approach to small molecules for the inhibition of this therapeutically important PPI. However, due to their polar nature and the negative net charge required for binding to Keap1, the peptides reported to date exhibit either mid-micromolar activity or are inactive in cells. Herein, we present a two-component peptide stapling strategy to rapidly access a variety of constrained and functionalised peptides that target the Nrf2/Keap1 PPI. The most promising peptide, P8-H containing a fatty acid tag, binds to Keap1 with nanomolar affinity and is effective at inducing transcription of ARE genes in a human lung epithelial cell line at sub-micromolar concentration. Furthermore, crystallography of the peptide in complex with Keap1 yielded a high resolution X-ray structure, adding to the toolbox of structures available to develop cell-permeable peptidomimetic inhibitors.

High resolution mass spectrometry (HRMS): HRMS was carried out using a Waters LCT Premier ® Time of Flight (ToF) mass spectrometer or the ThermoFinnigan Orbitrap Classic mass spectrometer.
Reported mass values are within the error limits of ± 5 ppm mass units.ESI refers to the electrospray ionisation technique.
Analytical HPLC: Chromatographs were obtained on an Agilent 1260 Infinity ® using a reversed-phase Supelcosil ABZ+PLUS column (150 mm x 4.6 mm, 3 μm) eluting with a linear gradient system (solvent A: 0.05% (v/v) TFA in water, solvent B: 0.05% (v/v) TFA in MeCN) over 15 min, unless otherwise stated, at a flow rate of 1 mL/min.HPLC was monitored by UV absorbance at 220 and 254 nm.
Preparative HPLC: Preparative HPLC was carried out on an Agilent 1260 Infinity ® using a reversedphase Supelcosil ABZ+PLUS column (250 mm x 21.2 mm, 5 μm) eluting with a linear gradient system (solvent A: 0.1% (v/v) TFA in water, solvent B: 0.05% (v/v) TFA in MeCN) over 20 min at a flow rate of 20 mL/min.HPLC was monitored by UV absorbance at 220 and 254 nm.

Small molecules synthesis and characterization
Staples D, F and 5 were prepared according to the generic synthetic route illustrated below (Scheme S1)
Couplings were carried out by adding HATU (4 equiv) to a solution of the Fmoc-protected amino acid (4 equiv) in DMF (~0.4 M).After 10 seconds, DIPEA (8 equiv) was added to the mixture.This preactivated mixture was then added to the resin in DMF and shaken for 3 minutes.The coupling time was extended in the case of amide coupling when substrates other than natural amino acids were used (30 minutes).The side chain protecting groups used were: t Bu for Asp, Glu, Thr; Boc for Lys; Pbf for Arg; Trt for Asn, Gln.Fmoc-Lys(iddve)-OH was used for conjugation of stearic acid with Lys.
On-resin attachment of stearic acid via Lys was achieved by orthogonal deprotection of the Lys(ddve) with 5% NH2NH2 in DMF (2 x 10 minutes) followed by standard amide coupling as described before.
Completion of amide couplings and Fmoc deprotection was determined by a chloranil test, in which acetaldehyde (200 μL) and a saturated solution of chloranil in toluene (50 μL) were added to a small amount of resin swelled in CH2Cl2.After 10 seconds shaking at rt, no change in colour indicated complete coupling, whilst green colouration of the resin indicated presence of a free amine.
Incomplete couplings were submitted to a second round of coupling.
Side chain deprotection and cleavage from the resin was achieved with TFA containing 2.5% TIPS and 2.5% H2O for 3 hours at rt.In case of methionine and cysteine-containing peptides, cleavage was achieved with TFA containing 5% EDT, 5% H2O and 2.5% TIPS.After cleavage, the mixture was filtered through a sintered funnel, the beads washed with MeOH and the filtrate was concentrated under a stream of N2.The crude residue was triturated with cold Et2O before purification by preparative HPLC.
H was not triturated with Et2O due to its lipophilic nature.

Automated Fmoc solid-phase peptide synthesis
Automated SPPS was carried out on solid-phase using a Fmoc-protecting group strategy on a CEM Liberty Blue ® automated microwave peptide synthesiser.
Automated peptide synthesis was performed using Merck LL MHBA Rink Amide resin (0.51 mmol/g, 1 equiv).All peptide couplings were performed with Fmoc-protected amino acids (5 equiv), Oxyma pure (10 equiv) and DIC (5 equiv) in DMF.Arg was coupled using double couplings for 15 min each without microwave irradiation.All other amino acids were coupled with 25 W power at 75°C over 15 min.
Fmoc deprotection was achieved with a solution of 20% piperidine in DMF, using 45 W power at 75°C over 3 min.N-terminal capping, cleavage and HPLC purification of peptides were carried out as previously described for manual SPPS (general method 10).

DVP stapling reaction
The linear peptide (1 equiv.) was dissolved in 50 mM NaPi buffer pH 8.0 (containing 5% of DMF) to a final concentration of 2 mg/mL.The staple (1.1 equiv) was added and the reaction shaken at rt for 1 h.Upon completion as monitored by analytical HPLC, the crude reaction was lyophilised and purified on a preparatory HPLC to yield the cyclised peptide.The prepared tag2 was cleaved from the resin by abovementioned procedure for 3 h, followed by HPLC purification.

Biophysics Experimental
Surface Plasmon Resonance (SPR) Peptide affinities to the Kelch domain of Keap1 is determined in a direct binding assay using 8K surface plasmon resonance (SPR) biosensor (Cytiva) at 20°C.Immobilization of Keap1 Kelch domain on a CM5 sensor chip (Cytiva) is performed using standard amine coupling procedure.The surface is washed with 10mM NaOH, 1M NaCl before activated with EDC/NHS (Cytiva), followed by immobilization of human Keap1 Kelch domain (A321-T609) fused with N-terminal 6x His-tag 7 (in 10 mM MES, pH=6.4).
Finally, the surface is deactivated by Ethanolamine.Immobilization levels are typically 5000-6000 RU.
The reference spot is treated as described, omitting the injection of Keap1.Compound concentration series are injected over the immobilized protein in increasing concentrations using single cycle kinetics (SCK) in running buffer (10 mM HEPES, 150 mM NaCl, 0.05% Tween20, 1 mM TCEP, pH 7.4).A 1:1 Langmuir interaction model is fitted to the experimental traces, enabling determination of kon, koff and Kd.

Keap1-Nrf2 TR-FRET Assay
In the time-resolved Förster resonance energy transfer (TR-FRET) assay 8 the IC50 of inhibitors which are able to inhibit the interaction between the Keap1-Kelch domain and a Nrf2 derived ETGE peptide are determined.The TR-FRET is measured between a Tb-labelled mAb (Anti-6-His, Tb cryptate Gold, CisBio 61H12TLB) that binds to Keap 1 Kelch domain and d2-labelled SA-biotin-peptide.
To determine the ability of synthesized peptide to inhibit PPI between the Keap1 Kelch domain and Nrf2, 1nM purified human Keap1 Kelch domain (A321-T609) fused with N-terminal 6x His-tag, 5 nM biotinylated Nrf2 peptide (KKKKAFFAQLQLDEETGEFL, Genescript, USA), d2-labeled streptavidin (Revvity, USA), and 0.1 mM anti-His monoclonal antibody tagged with terbium cryptate (Revvity, USA), were incubated in the presence of varying concentrations of investigated peptides for 3 h at room temperature.

1
Figure ES_1 -a) Ligand interaction diagram of P3-F with the Keap1 Kelch domain (PDB: 8Q1Q ); b) Ligand-interaction diagram of P8-F with the Keap1 Kelch domain (PDB: 8Q1R).Legend in the figure.c) Overlay of the two peptides P3-F in orange and P8-F in green) showing key residues maintain the same conformation whilst the solvent expose staples adopt a different conformation

Table S1 -
Characterisation of the peptides presented in this work.*Measured on a 5-95% gradient

Table S2 -
HPLC trace of peptides Details from the data processing and refinement are reported in tableS1.Pictures were generated with the software PyMol 17 .

Table S3 -
Statistics from crystallographic data reduction and refinement.Values in parenthesis refers to the highest resolution shell