A Ligand Selection Strategy Identifies Chemical Probes Targeting the Proteases of SARS‐CoV‐2

Abstract Activity‐based probes are valuable tools for chemical biology. However, finding probes that specifically target the active site of an enzyme remains a challenging task. Herein, we present a ligand selection strategy that allows to rapidly tailor electrophilic probes to a target of choice and showcase its application for the two cysteine proteases of SARS‐CoV‐2 as proof of concept. The resulting probes were specific for the active site labeling of 3CLpro and PLpro with sufficient selectivity in a live cell model as well as in the background of a native human proteome. Exploiting the probes as tools for competitive profiling of a natural product library identified salvianolic acid derivatives as promising 3CLpro inhibitors. We anticipate that our ligand selection strategy will be useful to rapidly develop customized probes and discover inhibitors for a wide range of target proteins also beyond corona virus proteases.


Synthesis of LS-Probe 2-tert-Butoxycarbonylamino-3-[4-(prop-2-ynyloxy)phenyl]-propionic acid propargyl ester (1)
The reaction was adapted from a protocol described in the literature 1 . 6 g of (21.3 mmol, 1 eq) N-tert-Butoxycarbonyl-tyrosine and 9 g of (56.1 mmol, 3 eq) K2CO3 were suspended in 30 mL dry DMF under N2 flow. After stirring for 10 min at room temperature 7.9 mL (73.1 mmol, 3.5 eq) of an 80 % solution propargyl bromide in toluene was slowly added. The solution was left to react for 18 h at room temperature. 150 mL H2O were used to quench the reaction. The mixture was extracted with diethyl ether, washed with distilled water and brine. The combined organic layers were dried over MgSO4 before solvent evaporation in vacuo. The yellow oil was used in the next step without any further purification (7.3 g, 100 %). Rf = 0.63 (petrol ether/ethyl acetate 1:1).

2-Amino-3-[4-(prop-2-ynyloxy)phenyl]-propionic acid propargyl ester (1b)
The reaction was adapted from a protocol described in the literature 1 . To 180 mL of MeOH on an ice bath, 21 mL (294 mmol, 15 eq) acetyl chloride were slowly added. The solution was left to stir for 10 min at 0 °C. 7 g of (1) were added and the solution was allowed to warm to room temperature. After 2 h the solvent was evaporated in vacuo to give the pure product as slightly brownish-white powder (4.1 g, 100 %).

O-Propargyl tyrosine (2)
The reaction was adapted from a protocol described in the literature 1 . To a mixture of 30 mL MeOH and 42 mL 2M NaOH 5.6 g (20 mmol, 1 eq) (1b) was added. The reaction was stirred for 17 h at room temperature. With concentrated HCl the pH of the mixture was adjusted to 7 and it was kept at 4 °C for 4 h. The precipitate was filtered off and dried in vacuo to give the pure product as light yellow powder (3.51 g, 17.1 mmol, 83 %).

2-tert-Butoxycarbonylamino-3-[4-(prop-2-yn-1-yloxy)phenyl]propanoic acid (5)
6.0 g Boc-ʟ-tyrosine (21.3 mmol) and 9.0 g K2CO3 (65.2 mmol, 3 eq.) were dissolved in 30 mL dry DMF. 6.4 mL propargyl bromide (97%, 82.4 mmol, 3.9 eq.) was added and the reaction stirred at room temperature for 20h. Water was added and the mixture extracted with ether, washed with brine, dried with Na2SO4, filtered and the solvent evaporated. TLC showed one product with only traces of impurities (Rf = 0.38 (petrol ether/ethyl acetate 5:1)). The yellow oil was dissolved in 30 mL methanol and 42 mL 2 M NaOH solution was added. The mixture was stirred at room temperature for 2h, adjusted to pH = 4-5 with 3 M HCl solution and extracted with ethyl acetate. The organic phase was washed with brine, dried with Na2SO4, filtered and the solvent evaporated leaving the product (5) as yellow solid (m = 6.06 g, 89 %). The product was used in the next step without further purification.

Plasmid preparation
The gene coding for the herein used proteins (t3CL pro , PL pro and their mutants) were custom synthesized and constructed in a pET-51b(+) plasmid (GenScript, New Jersey) for protein expression in E. coli BL21 cells. The expression vector was IPTG inducible with ampicillin marker and N-terminal Strep-tag II with cloning site KpnI-BamHI.

Transformation of plasmids
For preparation of competent cells, 1 mL of an overnight culture of E. coli BL21 cells was inoculated in 25 mL of LB medium and kept at 37 °C and 180 rpm. At an OD600 of 0.5 the cells were transferred into a 50 mL tarson tube and kept on ice for 20 min. Centrifugation at 6000 rpm and 4 °C for 10 min was performed. The supernatant was discarded and the cell pellet resuspended into 20 mL ice cold calcium chloride (50 mM). After an incubation time of 20 min on ice the cells were centrifuged at 6000 rpm for 10 min at 4 °C and the supernatant discarded. The cell pellet was again resuspended in 4.25 mL of calcium chloride solution (50 mM) as well as 0.75 mL of glycerol. Aliquots of 200 µL were made and directly put into liquid nitrogen before storing them at -80 °C.
An aliquot of competent E. coli BL21 cells was thawed on ice. 1 μL of the corresponding plasmid (50 ng/mL) was added to the cells and mixed by tabbing the tube three times. After incubation on ice for 30 min the cells were heat shocked at 42 °C for 45 sec. Afterwards the tube was placed on ice for 3 min. 900 μL LB media was added to the cells and they were kept at 37 °C for 1 h whilst shaking at 550 rpm. Centrifugation at 8000 rpm and 4 °C for 2 min was performed. The supernatant was discarded and the cell pellet resuspended. The cell suspension was streaked on ampicillin plates (100 µg/L), which were kept at 37 °C overnight. The next day colonies could be picked for overnight cultures which could be used to prepare glycerol stocks.

Overproduction of proteins
Overnight cultures for cellular assays were prepared by taking a small amount of a bacterial cryo-stock (15 % glycerol, stored at -80 °C) and inoculating them in 5 mL LB in sterile 13 mL polypropylene tubes (Sarstedt, ref 62.515.028), supplemented with antibiotics as indicated and grown for 14-16 h at 37 °C (180 rpm). 1 mL of a corresponding overnight culture was inoculated in 50 mL LB medium containing 100 μL/mL ampicillin and kept at 37 °C and 180 rpm. At an OD600 of 0.3 the protein expression was induced by adding 0.2 μg/mL of an IPTG solution (1 M). The cells were incubated at 37 °C and 180 rpm for 2 h before centrifugation at 4000 rpm for 20 min at 4 °C.

Affinity-purification of proteins
For purifying the proteins overexpression was performed in a scale of 3 L as described. The cell culture was centrifuged down at 4000 rpm for 20 min at 4 °C. The supernatant was discarded and the cell pellet washed with 15 mL PBS before centrifuging at 4000 rpm for 20 min at 4 °C. The cells were lysed by ultrasound treatment (25 % amplitude, 0.5 s ON, 2.1 s OFF, 20 pulses, Branson Digital Sonifier). Afterwards, samples were centrifuged for 1 h at 4000 rpm and 4 °C. The supernatant was transferred into a new 50 mL flask and put on ice before performing affinity purification via Strep-tag II on an ÄKTA start (GE Healthcare) using StrepTrap HP columns (GE Healthcare). Standard Bradford protocols were used to calculate the concentration of the protein fractions. 200 µL Aliquots of 0.4 mg/mL protein were frozen in liquid nitrogen before storing them at -80 °C.

LS-probe modification with ligands
In a 1.5 mL micro reaction tube were added 2.5 μL of 5 M pyridine (in DMSO), 2.5 μL of a probe stock (1 mM in butyl acetate) as well as 2.5 μL of the corresponding amine containing ligand (1 mM in DMSO) to result in a final concentration of 50 μM in the cell suspension. The mixture was incubated for 1 h at room temperature. Quenching was performed by adding 50 μL butyl acetate to the reaction mixture and pipetting the entire solution into a fresh 1.5 mL micro reaction tube containing 5 mg of (aminomethyl)polystyrene beads (70-90 mesh, Sigma-Aldrich). After incubation for 15 min the supernatant was transferred into a fresh 1.5 mL micro reaction tube. The beads were washed with 50 μL of butyl acetate and the supernatants were combined. The pooled solution was dried at high vacuum. 2 μL of DMSO were used to dissolve the reacted probe. For dose down experiments the same procedure as for the reaction of a LS-probe and ligand was used. To get a dose down of a final concentration of 50 μM, 20 μM, 10 μM, 5 μM, 1 μM and 0.1 μM in 50 μL the respective amount of ligand and probe was used.

In situ probe labelling of proteins in live E. coli cells
After overexpressing a protein in E. coli, for each sample 1 mL of the induced bacterial culture was transferred into a 1.5 mL Eppendorf tube. The cells were pelleted by centrifugation (4000 rpm, 7 min, 4 °C), washed with 50 μL PBS before re-suspending them in 48 μL PBS. The reacted and quenched probes, dissolved in 2 μL DMSO were added to the cell suspension before incubation at 400 rpm at 37 °C for 1 h. After incubation the cells were pelleted by centrifugation (4000 rpm, 5 min, 4 °C) washed with 50 μL PBS and resuspended in 50 μL PBS. The cell suspension was stored at -80 °C before further processing. After thawing the samples, they were lysed by ultrasound treatment (10 % amplitude, 0.5 s ON, 1 s OFF, 10 pulses, Branson Digital Sonifier). The resulting lysates were used for click chemistry and SDS-PAGE. After fluorescence scanning, Coomassie staining was applied to compare protein concentrations in the gel and validate the experiments.

In vitro probe labelling of proteins in lysates and purified protein
The respective protein aliquot (0.4 mg/mL) was thawed on ice. Reaction of LS probe and ligands was done as described. The reacted probe was dissolved in 2 μL DMSO and added with 8 μL PBS to 10 μL protein solution before incubation for 30 min at 400 rpm and 37 °C. After incubation Click Chemistry was performed.
The cell pellets were resuspended in 10 mL PBS buffer and lysed via sonification using a Branson Digital Sonifier (25 % amplitude, 0.5 s ON, 2.1 s OFF, 20 pulses). After centrifugation for 20 min at 9000 rpm and 4 °C, the supernatant was frozen immediately in liquid N2 and stored in aliquots at -80 °C.

In situ competitive profiling
The respective protein was overproduced in E. coli BL21 cells as described before. Reaction of LS probe and ligand was performed as described with 0.25 μL of LS-probe (1 mM), ligand (1 mM) and 5 M pyridine per sample. In the meantime, 1 μL of competitive molecule (10 mM) was incubated with 10 μL of the protein aliquot and 49 μL cell suspension for 1h at 25 °C and 400 rpm. Each sample of the reacted probe was dissolved in 0.25 μL DMSO and added to the cell suspension before incubation for 1 h at 400 rpm and 37 °C. The cells were further treated like described for in situ labeling with LS-probes.

In vitro competitive profiling
The respective protein aliquot (0.4 mg/mL) was thawed on ice. 10 μL of protein were used per reaction and pipetted into a 1.5 mL micro reaction tube together with 9.6 μL of PBS as well as 0.4 μL of the competitive compound (10 mM). The proteins were incubated for 30 min at 25 °C and 400 rpm. Reaction of probe and ligand was done as described with 0.1 μL LS-probe (1 mM), ligand (1mM) and pyridine (5 M) per sample. The reacted probe was dissolved in 0.1 μL DMSO and pipetted to the proteins and incubated for 30 min at 37 °C and 400 rpm. Afterwards Click Chemistry was performed directly.
For dose down experiments the same procedure as for in vitro competitive profiling with LS-probes was performed. Final concentrations of 200 μM, 100 μM, 50 μM, 25 μM, 10 μM, 5 μM and 1 μM of the competitive compound were used with 5 μM of the reacted probe.

SDS-PAGE and in-gel fluorescence scanning
Before performing SDS-PAGE the samples were incubated for 10 min at 95 °C and subsequently centrifuged down. SDS gels containing 10 % acrylamide and an aqueous solution of 37.5:1 acrylamide and N,N'-methylenbisacrylamide were used with a PeqLab system and run at 75 mA per gel. Visualization was done by in-gel fluorescence scanning (Fusion-FX7). Equal protein content and separation in SDS-gels was confirmed by Coomassie staining (InstantBlue TM , expedeon).

Proteomic experiments
In situ labelling was performed without doing click reaction subsequently. 2x SDS loading buffer was added directly to the resulting lysates. After SDS-PAGE and Coomassie staining the appropriate band was cut out of the gel and stored in a 1.5 mL micro reaction tube at 4 °C. Samples have been desalted using Nanosep 3K OMEGA centrifugal devices (Pall), digested with pepsin and measured on a Thermo LTQ Orbitrap Discovery with Eksigent 2D-nano HPLC (coverage of 84.4 %).

Substrate-based enzyme inhibition assays
All enzyme inhibition assays were performed in triplicates in black, flat-bottom 96-well plates with a total volume of 100 µL. PL pro : Purified recombinant enzyme was diluted to a concentration of 200 nM in PBS and 5 µL inhibitor dissolved in DMSO were added to the desired concentrations (250 µM, 100 µM, 50 µM, 25 µM, 10 µM, 2.5 µM, 1 µM, 0.5 µM). A positive control with 5 µL DMSO and a negative control without enzyme were performed alongside every measurement. After incubation for 10 min at room temperature, 5 µL of 2.5 mM fluorogenic substrate in DMSO (Arg-Leu-Arg-Gly-Gly-AMC, >98 % purity, Thermo Fisher) were added and fluorescence (λex = 360 nm, λem = 460 nm) was measured at 37 °C every 30 s over 15 min in a Tecan M200 infinite Pro plate reader with 10 s of vigorous shaking before each measurement.

IC50 values:
For each concentration of inhibitor, the enzymatic turnover rate [1/s] was determined by linear regression, whereby the first 5 min of fluorescence measurement were excluded for the data sets with PL pro . Average percentages of enzyme activity and corresponding standard deviations were calculated and plotted against the logarithm of inhibitor concentration. Subsequent sigmoidal fit using GraphPad Prism gave the respective IC50 values.

Preparation of root extracts from red sage (Salvia miltiorrhiza)
For preparation of crude extracts, a protocol for purification of salvianolic acids was followed 4 . 10 g commercially available Salvia miltiorrhiza root powder was suspended in 200 mL 60 % ethanol (20 mL per 1 g powdered root) and subjected to sonication for 1 h. The mixture was filtered and the flowthrough was concentrated in vacuo below 40 °C. Subsequent lyophilization yielded 4.916 g of crude extract as a brown solid. To further concentrate the extract, 4.5 g crude extract were resuspended in 50 mL H2O and acidified with HCl to pH = 2. The mixture was extracted five times with ethyl acetate, the combined organic phases were dried over MgSO4 and the solvent was evaporated, yielding 0.535 g of a dark red solid.

Molecular docking
Molecular docking was carried out with AutoDock 4.2.6 and AutoDock Tools 1.5.6 5 . As the Ligands are covalently bound we applied the 'flexible side chain' covalent docking as described by Bianco et. al. 6 SARS-CoV-2 main protease (PDB ID: 6YB7) structure was retrieved from the RCSB PDB database (www.rcsb.org). All files required for docking of proteins and ligands were prepared by adapted "addcovalent" scipts provided by the authors of the covalent docking method utilizing AutoDock Tools 1.5.6 (ADT). The receptor was obtained from the PDB file by removing all waters and co-factors, adding hydrogens and calculating Gasteiger charges. All synthesized compounds were drawn in Chem Draw Professional 17.0 with the CS atoms of covalently bound cysteine and exported as smiles code. Maestro (Schrödinger Release 2019-2: Maestro, Schrödinger, LLC, New York, NY, 2020) was used to generate 3D structures from smiles and perform structure minimization/optimization with the OPLS 2005 force field 7 . In the flexible side chain method, two receptor atoms added to the ligand coordinates are used to superimpose the ligand on the appropriate residue in the target protein. ADT was used to add hydrogens, calculate Gasteiger charges, and generate a modified flexible ligand, using default methods. During docking the resulting side chain-ligand structure is treated as flexible, allowing optimization of the interaction of the tethered ligand with the rest of the protein. Grid maps were calculated following the standard AutoDock protocol for flexible side chains. The Lamarckian Genetic Algorithm was used with default settings, generating 10 poses. 5 independent docking simulations have been performed for every ligand.
The structural models of the N-terminal 3CL pro -Strep-tag construct (t3CL pro ) and the corresponding native protein before autocleavage have been modeled with the comparative modeling software MODELER 8 by extending the SARS-CoV-2 main protease dimer structure (PDB ID: 6Y2E 9 ) by the Nterminal part. 50 models have been constructed for every protein by applying the adjusted automodel scheme where only the new N-terminal part with 3 connected residues of the x-ray structure was optimized in order to stay as close as possible to the reference structure 6Y2E. The 10 best DOPE (Discrete Optimized Protein Energy) 10 scored models are then used to identify the covered part of the binding pocket of SARS-CoV-2 main protease.

Tables
Supporting Table S1 Overview of the amine ligand library. (continued on following pages)

Ligand
Ligand

Fig. S3
Diversification reaction of the LS-probe with the ligand 08 (primary aliphatic amine) with pyridine is completed after 15 min as indicated by 1 H NMR spectroscopy in DMSO-d6. a, overlaid spectra of the reaction (gold) and the unreacted LS-probe (blue). b, overlaid spectra of the reaction (turquoise) and the unreacted ligand (brown).

Fig. S4
Examples of gels for the in situ labelling of t3CL pro and PL pro by the LS probes. a, initial screening for active probes against t3CL pro . b, Comparison of wild type (wt) and mutants (m) for t3CL pro . c, Comparison of wild type (wt) and mutants (m) for PL pro .