Targeting the Main Protease (Mpro, nsp5) by Growth of Fragment Scaffolds Exploiting Structure-Based Methodologies

The main protease Mpro, nsp5, of SARS-CoV-2 (SCoV2) is one of its most attractive drug targets. Here, we report primary screening data using nuclear magnetic resonance spectroscopy (NMR) of four different libraries and detailed follow-up synthesis on the promising uracil-containing fragment Z604 derived from these libraries. Z604 shows time-dependent binding. Its inhibitory effect is sensitive to reducing conditions. Starting with Z604, we synthesized and characterized 13 compounds designed by fragment growth strategies. Each compound was characterized by NMR and/or activity assays to investigate their interaction with Mpro. These investigations resulted in the four-armed compound 35b that binds directly to Mpro. 35b could be cocrystallized with Mpro revealing its noncovalent binding mode, which fills all four active site subpockets. Herein, we describe the NMR-derived fragment-to-hit pipeline and its application for the development of promising starting points for inhibitors of the main protease of SCoV2.


Screening data
Screening data with the 768 compounds involving DSI-poised library purchased from Enamine were conducted as previously described. 1NMR screening against the DSI-poised library purchased from Enamine with 896 compounds employed 10 mM phosphate buffer (pH 7.6) containing 50 mM NaCl in D2O, 0.04% NaN3 and 3.85% d6-DMSO at 298 K.The protein concentration was 11.5 µM and mixtures of max.6 compounds with 320 µM each were added.STD experiments were recorded, on a 600 MHz Bruker spectrometer with cryoprobe, with irradiation for 3 s at 200 Hz power using trains of 50 ms gaussian pulses, number of scans 256, acquisition time 0.98 s, relaxation delay 2 s.The protein signal was filtered using a spinlock for 20 ms at a 10 kHz power level.S/N > 10 was taken as a hit, everything below was considered a negative.The screening against the Bionet library containing 428 compounds and the Maybridge library containing 800 compounds were performed on a Bruker 600 MHz Avance III HD spectrometer equipped with a 5 mm BBO cryogenic probe.In addition, the Bionet library was also screened using a Bruker 700 MHz equipped with a cryogenic QCI-F probe.Screening conditions were very similar for both libraries, using a 10 mM phosphate buffer (pH 7.4) containing 130 mM NaCl, 2.4 mM KCl, 10% D2O and 8% (Bionet) or 1% (Maybridge) d6-DMSO.The protein concentration was 10 µM and 9-11 ligands were added to a final concentration of 100 µM for each ligand.Samples (180 µL final volume) were transferred to 3 mm tubes.STD experiments were recorded using a pulse-train of 40 gauss pulses (50 ms each) for the on-and off-resonance irradiation with a field strength of 200 Hz.The length of the CPMG spin-lock in the T2-experiments was 200 ms and the spin-lock in the pseudo 2D T1rho-experiments were 10 ms and 200 ms with a field strength of 5 kHz.For the Bionet, 19 F T2-experiments were recorded library was also screened using 19 F NMR spin-echo experiments with a total spin-echo time of 100 ms and with a 70 kHz Chirp inversion pulse. 2

Crystallization Protein expression and purification
Nsp5 for crystallization was prepared as follows: The nsp5 gene was cloned into a pGEX-6-1 vsector with a C-terminal His6-tag by the Hilgenfeld group. 3The provided plasmid was transformed into E. coli strain BL21-Star (DE3).An overnight culture was inoculated from transformed clones and transferred the next day into auto-induction medium. 4The cells were grown at 37°C until OD = 0.6 and then temperature was lowered to 18°C for overnight expression.Cells were harvested the next day through centrifugation at 10.000xg for 10 min.Cell pellets were flash-frozen in liquid nitrogen until purification.For purification the cells were thawed and resuspended in 20 mM Tris (pH 7.8), 150 mM NaCl, 5 mM imidazole, 0.05% Tween.The cells were lysed via sonication and centrifuged at 40.000x g for 45 min at 4°C.The supernatant was loaded onto a Ni-NTA affinity chromatography column, and the protein was eluted with a stepwise imidazole gradient up to 500 mM imidazole.The protein was dialyzed overnight with addition of 1:10 w/w PreScission into 20 mM Tris (pH 7.8), 150 mM NaCl, 1 mM DTT.The next day the cleaved protein was subjected to a Ni-NTA affinity chromatography column to remove the cleaved His6-tag.The final purification step was size exclusion chromatography using a Superdex 75 column equilibrated in 20 mM Tris (pH 7.8), 150 mM NaCl, 1 mM TCEP, 1 mM EDTA.The fractions with the purified protein were pooled, concentrated to 5 mg/ml and flash-frozen in liquid nitrogen before storage at -80°C.

Crystallization conditions
Nsp5 was crystallized in 23.5% PEG 1.500, 0.2 M MIB (pH 7.4), 5% DMSO, 0.025 mM EDTA (pH 7.0) using the NT8 pipetting robot (Formulatrix) and MRC 3-lens 96-well low-profile plates.The final drop consisted of 200 nl protein, 100 nl reservoir and 50 nl 1:50 seed stock dilution and was equilibrated at 20°C against 40 µl reservoir.Initial seeds were kindly provided by Deniz Eris (PSI, Villigen, Switzerland) and predetermined the crystallization in the orthorhombic space group.Based on the crystals grown from the initial seeds, new seeds were prepared in the following way.Crystals from one drop were crushed and transferred into 50 µl reservoir solution.The mixture was vortexed four times for 30 s with 30 s on ice in between with a Seed Bead™ (Hampton Research).The final seed stock was diluted to 1:50, aliquoted and flash-frozen in liquid nitrogen.Crystals grew within 2 days.The soaking solution was the same as the crystallization solution except for an additional 5 mM of the respective compound.A soaking plate was prepared with 40 µl crystallization solution as reservoir and a 0.4 µl drop of the respective soaking solution.The crystals were transferred from the crystallization plate into the soaking drops.After the transfer the plate was sealed with crystallization foil and incubated overnight at 20°C.The next day the crystals from each soaking drop were harvested, flash-frozen in liquid nitrogen and stored until measurement.
Data collection was performed at BL14.1 (BESSY II, Berlin, Germany). 5The data was collected with 2000 images in 0.1° increments, an exposure time of 0.08 s, a 100 µm aperture and at 13.5 keV.The collected data was then processed automatically via XDSAPP 6 and afterwards automatically refined with fspipeline. 7The input model for fspipeline was an inhouse refined MPro model, based on a molecular replacement with the PDB model 7BB2.For the dataset of 37b the binding could be confirmed in the electron densitity and the ligand could be fitted into the electron density manually after the fspipeline run.The resulting model was then further subjected to iterative cycles of manual building in coot 8 and refinement using phenix.refine. 9he nsp5-37b co-crystal structure is deposited at the PDB with the ID 8PH4.The table with data collection and refinements statistics are given in table S3.S1 for all fragments screened).The numbers in the structures represent the manuscript IDs of the 29 primary screening hits of four libraries.
Pre-incubation with 200 µM nsp5 in 50 mM NaPi (pH 7.0), without/with 0.1-2 mM ligand, without/wit h 4% DMSO (present for all ligands and the DMSO reference), without/with 1 mM TCEP (present only for the TCEP reference) was at 22°C for 1 h or 24 h.The pre-incubation time is indicated in the figure if both times are tested.
Cleavage reactions are started by the addition of substrate FLAG3-His6-SAVLQ-nsp9 (235-296 µM) diluting nsp5 to 100 µM, ligand to 0.05-1 mM, DMSO to 2%, while the buffer stayed the same.The cleavage time is indicated in the figure by a bar with the starting and end point.
All cleavage reactions were analyzed by SDS-PAGE and for all a protein marker (M, PageRuler™ Unstained Low Range Protein Ladder (Thermo Scientific™)), nsp5 (Ref.), and FLAG3-His6-SAVLQ-ns p9 (Ref.) were used.The gels in (a) are exemplary shown in full showing the marker and nsp5.All other gels are cut to only show full-length substrate and cleaved nsp9.All protein/peptide migration levels are indicated on the right of the gels.Gels are labeled as +TCEP for nsp5 with TCEP; -TCEP, +DMSO for nsp5 with DMSO; and -TCEP for nsp5 without reducing agent; or with the corresponding ligand.

Synthesis of Z604 and its derivatives
Synthesis of secondary and tertiary amines based on the scaffold of Z604 [1]  [2] Scheme S1: Reaction to synthesize secondary [1] and their corresponding tertiary amines [2].[1] The general reaction procedure was based on a synthesis known from literature. 16,17 muracil SI1 (1 eq) and amine SI2a-d (1 eq) reacted without solvent at 160 °C in the melt for 1-3 h.Dest.water was added to the reaction mixture and the pH was adjusted to neutral with 10% HCl.In 30a and 30b solids precipitated, these were filtered off and characterized as clean product.For product 30c, the whole reaction mixture was adsorbed on Celite and purified by reversed phase flash chromatography (H2O→MeCN).Products 30a-d were clearly identified by NMR spectroscopy and mass spectrometry.The spectra and chromatograms can be found in the appendix.
[2] The general reaction procedure was based on a synthesis known from literature. 18e secondary amine 30a-d (1.0 eq) was dissolved in 5 mL dry DMSO and Bromomethylpyridine (1.1 eq) and DIPEA (5.0 eq) were added under Schlenk conditions.The reaction mixture was stirred for 1-3 days at room temperature.The solvent was removed under reduced pressure and the purification was done via reversed phase flash chromatography (H2O→MeCN) following HPLC.Products 31a-d were clearly identified by NMR spectroscopy and mass spectrometry.The spectra and chromatograms can be found in the appendix.To obtain SI10, SI8 (1 eq), EDC•HCl (1 eq) and HOBt (1 eq) were dissolved in 5 mL abs.DMF at 0 °C.SI9 (1 eq) was added and the reaction mixture was stirred for 15 min at 0 °C, heated up to room temperature and stirred for 1 h.For purification, the reaction mixture was diluted with EtOAc, then washed three times each with 20 mL of 1 M HCl, NaHCO3 solution and NaCl solution.The organic layers were combined and the solvent was removed under reduced pressure and purified by reverse-phase flash chromatography (Celite, H2O→ACN).Product SI10 was clearly identified by NMR spectroscopy and mass spectrometry.
To obtain products 32a-b, compound SI10 (1 eq) and SI11/30a (3 eq) were dissolved in 2-3 mL abs.DMF.The reaction mixture was stirred at 160 °C for 3 h, cooled to room temperature and stirred for 1 h.The solvent was removed under reduced pressure and the crude product purified by reversed-phase flash chromatography (Celite, H2O→ACN).Products 32a-b were clearly identified by NMR spectroscopy and mass spectrometry.The spectra and chromatograms can be found in the appendix.Reactants SI12 (1 eq), SI13a-b (1 eq) and SI14 (1 eq) were dissolved in 6.5 mL abs.DMF.

Synthesis of X77-like compounds containing Uracil moiety
The mixture was cooled to 0 °C in an ice bath, reactant SI15 (1 eq) was added and stirred for 1 h.Additionally the reaction mixture was stirred overnight at room temperature.For purification, the reaction mixture was extracted with DCM and washed with water and brine.

S23
After combining the organic phases, the solvent was removed under reduced pressure.The oily crude product was purified by reversed-phase flash chromatography (Celite, H2O→MeCN).The selected isolated fractions were combined and the solvent was removed under reduced pressure.A yellow solid was obtained, which was additionally purified by normal phase flash chromatography (silica gel, DCM → DCM/MeOH 4:1).Products 35a-b were clearly identified by NMR spectroscopy and mass spectrometry.The spectra and chromatograms can be found in the appendix.

Figure S1 :
Figure S1: Docking poses of primary screening hits 1-29 generated by SwissDock 10, 11 using the nsp5 crystal structure PDB 5R83 (monomeric) showing the active site pocket (s.TableS1for all fragments screened).The numbers in the structures represent the manuscript IDs of the 29 primary screening hits of four libraries.

Figure S2 :
Figure S2: NMR spectra (1D 1 H, waterLOGSY, and T2-CPMG (5 ms and 100 ms)) of commercially acquired Z604 in a mixture of 12 fragments in presence of GHMnsp5.A single fragment spectrum is used for chemical shift deconvolution in the mixture.Z604 shows sign changes in the waterLOGSY in presence of GHMnsp5.At the start of this project, we classified this compound as a potential hit after confirmation of its binding properties by other biophysical methods (SPR, TSA).The sample was prepared with 10 µM GHMnsp5, 200 µM of each ligand in 25 mM NaPi (pH 7.5), 150 mM NaCl, 5% d6-DMSO.

Figure S3 :
FigureS3: Properties of the primary screening hits of all four libraries.
FigureS4: 1D-1 H spectra of 0.25 mM fresh (a-c) or aged (d-f) Z604 in 50 mM NaPi (pH 7.0), 10% D2O, 0.25% d6-DMSO in absence of reducing agent, in presence of 1 mM TCEP or 1 mM DTT.Both TCEP and DTT are faster oxidized in presence of Z604.In case of aged Z604 this process is accelerated.A new signal in the aromatic region (7.12 ppm) appears in presence of TCEP.The solubility/stability of aged Z604 is better in presence of reducing agent.

Figure S5 :
Figure S5: 1D-1 H spectra of Z604 (a), and sections of the aliphatic (b), and aromatic regions (c).A stock solution of 100 mM Z604 in DMSO is prepared and stored at rt to capture the aging.For each time point (0-62 d) a new NMR sample of 0.25 mM Z604 in 50 mM NaPi (pH 7.0), 10% D2O, 0.25% d6-DMSO is prepared and directly measured.During the aging at rt the Z604 stock changes its color from colorless to pink/red/brown.A second signal set appears with time and gets more intense.After 62 d the second signal set yields ~92%.The same NMR samples are measured after a few days and the second set of signals disappears hinting at instability or limited solubility in buffer.Data not shown but visible in SIfig.4(e).The second signal set disappears after 20 h in absence of reducing agent (red and blue spectra).

Figure
Figure S6: [ 15 N, 1 H]-BEST-TROSY of wt nsp5 with backbone assignment to a degree of 83% (86% when leaving out proline residues and the N-terminal Ser).(a) shows the whole spectrum, (b) and (c) show sections, which are marked by color in the full spectrum.

Figure S10 :
Figure S10: Sections of overlayed [ 15 N, 1 H]-BEST-TROSY spectra of apo nsp5 (black) and nsp5 with ligands (color coding on the bottom of figure) showing amino acids V42 and G174 that are affected by the ligand presence (full spectra in Figure S1).

Figure S11 :
Figure S11: Modeling of each ligand within the active site pocket of nsp5 including amino acids N142 and Q189 (upper row: non-covalent, lower row: covalent binding of ligand)

Figure S13 :
Figure S13: The co-crystal structure of nsp5 with 35b as a dimer (a).Tilted view to show the second binding site (b) and a section (c).The light tale colored amino acids show CSPs > 0.02, and tale colored amino acids show CSPs > 0.05 at a 10-fold excess of ligand over protein (s. Figure S7) supporting the allosteric binding site.

Table S3 :
Crystal structure data collection and refinement statistics .