key: cord-0888875-3ls4hexf
authors: Kitamura, Naoya; Sacco, Michael Dominic; Ma, Chunlong; Hu, Yanmei; Townsend, Julia Alma; Meng, Xiangzhi; Zhang, Fushun; Zhang, Xiujun; Kukuljac, Adis; Marty, Michael Thomas; Schultz, David; Cherry, Sara; Xiang, Yan; Chen, Yu; Wang, Jun
title: An expedited approach towards the rationale design of non-covalent SARS-CoV-2 main protease inhibitors with in vitro antiviral activity
date: 2020-12-20
journal: bioRxiv
DOI: 10.1101/2020.12.19.423537
sha: ddaa16feef4403068950931c548066e10558c766
doc_id: 888875
cord_uid: 3ls4hexf

The main protease (Mpro) of SARS-CoV-2 is a validated antiviral drug target. Several Mpro inhibitors have been reported with potent enzymatic inhibition and cellular antiviral activity, including GC376, boceprevir, calpain inhibitors II and XII, each containing a reactive warhead that covalently modifies the catalytic Cys145. In this study, we report an expedited drug discovery approach by coupling structure-based design and Ugi four-component (Ugi-4CR) reaction methodology to the design of non-covalent Mpro inhibitors. The most potent compound 23R had cellular antiviral activity similar to covalent inhibitors such as GC376. Our designs were guided by overlaying the structure of SARS-CoV Mpro + ML188 (R), a non-covalent inhibitor derived from Ug-4CR, with the X-ray crystal structures of SARS-CoV-2 Mpro + calpain inhibitor XII/GC376/UAWJ247. Binding site analysis suggests a strategy of extending the P2 and P3 substitutions in ML188 (R) to achieve optimal shape complementary with SARS-CoV-2 Mpro. Lead optimization led to the discovery of 23R, which inhibits SARS-CoV-2 Mpro and SARS-CoV-2 viral replication with an IC50 of 0.31 μM and EC50 of 1.27 μM, respectively. The binding and specificity of 23R to SARS-CoV-2 Mpro were confirmed in a thermal shift assay and native mass spectrometry assay. The co-crystal structure of SARS-CoV-2 Mpro with 23R revealed the P2 biphenyl fits snuggly into the S2 pocket and the benzyl group in the α-methylbenzyl faces towards the core of the enzyme, occupying a previously unexplored binding site located in between the S2 and S4 pockets. Overall, this study revealed the most potent non-covalent SARS-CoV-2 Mpro inhibitors reported to date and a novel binding pocket that can be explored for Mpro inhibitor design.

by coupling structure-based drug design and the Ugi four-component reaction methodology to the design of non-covalent M pro inhibitors. Specifically, through a screening of a focused library of protease inhibitors, we recently discovered several non-canonical SARS-CoV-2 M pro inhibitors including boceprevir, and calpain inhibitors II and XII 7 . These inhibitors differ from classic M pro inhibitors such as GC376 in that their P1 substitution does not contain a glutamine mimetic. The co-crystal structures of calpain inhibitors II and XII with SARS-CoV-2 M pro revealed a critical hydrogen bond between the methionine side chain from calpain inhibitor II and pyridinyl substitution from calpain inhibitor XII with the H163 side chain imidazole located at the S1 pocket 8 . Similarly, the carbonyl from the pyrrolidone in GC376 also forms a hydrogen bond with the H163 side chain imidazole 7 . Given the importance of this hydrogen bond with H163 for the high affinity binding of inhibitors to SARS-CoV-2 M pro , we hypothesize that noncovalent inhibitors without a reactive warhead targeting the Cys145, but retain the hydrogen bond capacity with H163 can be designed as potent SARS-CoV-2 M pro inhibitors. In this study, we report the structure-based design of non-covalent M pro inhibitors based on the overlaying structures of SARS-CoV or SARS-CoV-2 M pro in complex with existing inhibitors or the peptide substrate. The design was based on the scaffold of ML188 (R) 12 , a non-covalent SARS-CoV M pro inhibitor, which similarly contains a pyridinyl in the P1 substitution as calpain inhibitor XII.

The overlaying structures revealed a strategy of extending the P2 and P3 substitutions in ML188 (R) to occupy the extra space in the S2 and S3/S4 pockets of SARS-CoV-2 M pro in order to increase the binding affinity. The most potent inhibitor from this study 23R showed enzymatic inhibition and cellular antiviral activity similar to the covalent inhibitor GC376. Its mechanism of action was characterized in the thermal shift-binding assay, native mass spectrometry binding assay, and enzyme kinetic studies. An X-ray crystal structure of SARS-CoV-2 M pro in complex with 23R was solved, revealing a previously unexplored binding site in between the S2 and S4

pockets. Overall, this study led to the identification of the most potent non-covalent M pro inhibitor 23R with potent enzymatic inhibition and in vitro cellular antiviral activity with a novel mechanism of action.

Among the non-canonical SARS-CoV-2 M pro inhibitors we recently discovered, calpain inhibitor XII has an unexpected binding mode showing an inverted conformation in the active site 8 (Fig. 1a) . Instead of projecting the norvaline and leucine side chains into the S1 and S2 pockets as one would expect from its chemical structure, the pyridinyl substitution snuggly fits in the S1 pocket and forms a hydrogen bond with the H163 imidazole (Fig. 1a) . This hydrogen bond is essential, as replacing the pyridine with benzene led to an analog UAWJ257 with a significant loss of enzymatic inhibition 8 . Examining the X-ray crystal structures of SARS-CoV and SARS-CoV-2 M pro in the PDB database revealed another compound ML188 (R) 12 , which shares a similar binding mode with calpain inhibitor XII. ML188 (R) is a non-covalent SARS-CoV M pro inhibitor derived from a high-throughput screening hit 12 . The pyridinyl from ML188 (R) similarly fits in the S1 pocket and forms a hydrogen bond with the H163 side chain imidazole ( Fig. 1b) . In addition, the furyl oxygen and its amide oxygen both form a hydrogen bond with the G143 main chain amide amine. ML188 (R) was reported to inhibit the SARS-CoV M pro with an IC 50 value of 1.5 ± 0.3 µM and the SARS-CoV viral replication in Vero E6 cells with an EC 50 value of 12.9 µM 12 . Several follow up studies have been conducted to optimize the enzymatic inhibition and cellular antiviral activity of this series of compounds, however, no significant improvement has been made 13, 14 . The similar binding mode of ML188 (R) with calpain inhibitor XII, coupled with the convenient synthesis through the one pot Ugi-4CR, inspired us to design non-covalent SARS-CoV-2 M pro inhibitors based on the ML188 (R) scaffold. Specifically, we leverage our understanding of the M pro inhibition mechanism based on the X-ray co-crystal structures of SARS-CoV-2 M pro with multiple inhibitors to guide the lead optimization (Figs. 2a-d) 7, 8 .

Overlaying the X-ray crystal structures of SARS-CoV M pro + ML188 (R) (PDB: 3V3M) and the SARS-CoV M pro H41A mutant + the peptide substrate (PDB: 2Q6G) revealed that the furyl, 4tert-butylphenyl, pyridinyl, and tert-butyl of ML188 (R) fit in the S1', S2, S1, and S3 pockets respectively (Figs. 2a and 2d) . Therefore, the furyl, 4-tert-butylphenyl, pyridinyl, and tert-butyl substitutions in ML188 (R) were defined as P1', P2, P1, and P3, respectively. Next, overlaying the structure of SARS-CoV M pro + ML188 (R) (PDB: 3V3M) and SARS-CoV-2 M pro + GC376 (PDB: 6WTT) suggested that the tert-butyl at the P3 substitution of ML188 (R) can be extended to fit in the S4 pocket (Figs. 2b and 2d). Previous structure-activity relationship studies of GC376 indicate that P4 substitution is important, while P3 substitution does not contribute significantly to the binding affinity, as it is solvent exposed 3, 8, 9, 15 . Similarly, the overlaying structures of SARS-CoV M pro + ML188 (R) (PDB: 3V3M) and SARS-CoV-2 M pro + UAWJ247 (PDB: 6XBH) suggested that the 4-tert-butyl at the P2 substitution of ML188 (R) can be replaced by phenyl to occupy the extra space in the S2 pocket (Figs. 2c and 2d). Overall, the design mainly focuses on extending the P2 and P3 substitutions of ML188 (R) to achieve optimal shape complementarity with the SARS-CoV-2 M pro (Fig. 2e) . In practice, we adopted a stepwise optimization procedure in which the P3 and P2 substitutions were optimized individually in step 1, and then the optimal P2/P3 substitutions were combined in step 2 (Fig. 2e) . Guided by the design rationale elucidated above, a focused library of ML188 analogs were designed and synthesized (Fig. 3) . As the P1' furyl and P1 pyridinyl both form a critical hydrogen bond with the M pro (Figs. 3a-b), the P1' and P1 substitutions were kept with minimal variations (Fig. 3c ). All designed compounds were synthesized using the one pot Ugi fourcomponent reaction and tested as enantiomer/diastereomer mixtures (Fig. 3c) . To circumvent the need of relying on expansive chiral HPLC column for the separation of enantiomers, we strategically introduced the chiral isocyanide so that the diastereomer product mixture can be separated by convenient silica gel column or reverse phase HPLC column purification. 

In total, 39 compounds were synthesized (Figs. 4a-4e) and all compounds were initially tested as a mixture of enantiomers or diastereomers in the FRET-based enzymatic assay against SARS-CoV-2 M pro at 20 µM (Fig. 4f) . Compounds showing more than 50% inhibition at 20 µM were further titrated to determine the IC 50 values. Next, compounds with IC 50 values lower than 5 µM were selected for cellular cytotoxicity profiling in Vero E6 cells, the cell line which was used for the SARS-CoV-2 antiviral assay. The purpose was to prioritize lead candidates for the in vitro cellular antiviral assay with infectious SARS-CoV-2. Compounds with potent enzymatic inhibition (IC 50 < 5 µM) but moderate to high cellular cytotoxicity (CC 50 < 100 µM) were labeled in red. Compounds with both potent enzymatic inhibition (IC 50 < 5 µM) and low cellular cytotoxicity (CC 50 > 100 µM) were labeled in blue (Figs. 4a-4e). As shown in Fig. 4f , the majority of the designed compounds showed more than 50% inhibition when tested at 20 µM.

Specifically, Fig. 4a lists compounds with P4 variations. As a reference, ML188 (1) (racemic mixture) inhibits SARS-CoV-2 M pro with an IC 50 value of 10.96 ± 1.58 µM. It was found that Replacing the tert-butyl in compound 13 with the bulkier trimethylsilyl led to compound 14 with a 2.9-fold increase in M pro inhibition. Cyclohexyl (17) , thienyl (19), pyrrolyl (20) , pyridinyl (21) , and phenyl (23) were found to be the most favorable substitutions at the S2 pocket. Compound 16 with piperidyl substitution had similar potency as compound 13, while compound 15 with O-tert butyl was less active. Further extending the substitution to benzyl led to compound 22 that was inactive, suggesting biphenyl might be the largest substitution that can be accommodated at the S2 pocket.

The P1' and P1 substitutions (Figs. 4c and 4d) were chosen to retain the critical hydrogen bonds in ML188 (Fig. 3a) . It was found that imidazole (24) was tolerated at the P1' position (IC 50 = 0.96 ± 0.09 µM), followed by isoxazole (25) Next, compounds with potent enzymatic inhibition (IC 50 ≤ 1 µM) and low cellular cytotoxicity (CC 50 > 100 µM) were prioritized for the cellular antiviral assay with infectious SARS-CoV-2 in Vero E6 cells using the immunofluorescence assay as the primary assay ( Given the potent antiviral activity and a high selectivity index of these potent lead compounds, we then selected compound 23 for further characterization. The two diastereomers of 23 were separated by reverse phase HPLC (Fig. 5 ). Both diastereomers were tested in the FRET-based enzymatic assay. GC376 was included as a positive control. It was found that 23R

is the active diastereomer with an IC 50 value of 0.31 ± 0.04 µM, while the 23S diastereomer was more than 16-fold less active (IC 50 = 5.61 ± 0.71 µM) ( Table 2 ). The stereochemistry of 23R was determined by the co-crystal structure with SARS-CoV-2 M pro as described in the following section. Compared with the parent compound ML188 (1), the optimized lead 23R had more than a 35-fold increase in enzymatic inhibition against SARS-CoV-2 M pro . Compound 23R also showed comparable potency against SARS-CoV M pro with an IC 50 value of 0.27 ± 0.03 µM.

Neither ML188 (1) nor 23R inhibited the SARS-CoV-2 papain-like protease (PL pro ) (IC 50 > 20 µM) ( Table 2) , suggesting the inhibition of SARS-CoV-2 M pro by 23R is specific. inhibited SARS-CoV-2 (USA-WA1/2020 isolate) replication in Calu-3 cells with an EC 50 value of 3.03 µM and it was not cytotoxic at up to 100 µM (Fig. 6) . The 2.4-fold difference in antiviral potency between the Vero E6 and Calu-3 cell lines might due to differences in cell membrane permeability or metabolism. 

The mechanism of action of 23R was characterized using the native mass spectrometry binding assay, the thermal shift binding assay, and the enzymatic kinetic studies (Fig. 7) . In the native mass spectrometry binding assay, compound 23R showed dose-dependent binding to SARS-CoV-2 M pro , similar to the positive control GC376, with a binding stoichiometry of one drug per monomer (Fig. 7a) . Similarly, compound 23R showed dose-dependent stabilization of the SARS-CoV-2 M pro in the thermal shift binding assay with an apparent K d value of 9.43 µM, a 9.3-fold increase compared to ML188 (1) (Fig. 7b) 

Using X-ray crystallography, we successfully determined the binding pose of 23R with SARS-CoV-2 M pro at 2.6 Å resolution (Fig. 8a) . Electron density reveals the body of 23R extends throughout the substrate binding channel, with side chains occupying the S1', S1, S2, and S3

sub-pockets. The binding pose is similar to the previously solved structure of SARS-CoV M pro with ML188 (R) (PDB: 3V3M) 12 , consistent with the similarities between the two compounds and between the two proteins (Fig. 8b) . The furyl moiety of 23R binds to a portion of the P1' site, which normally accommodates small hydrophobic residues. While the furylamide carbonyl group of 23R does not insert into the oxyanion hole, it does form a bifurcated hydrogen bond with the apical residue of this oxyanion hole, Gly143. However, the furan ring oxygen is likely a weaker hydrogen bond acceptor than the amide oxygen and it lies outside of the plane of Gly143's amide NH. Directly attached to the furylamide moiety is a P2 biphenyl group and a P1 pyridinyl ring. The P2 biphenyl group projects directly into the S2 pocket, which prefers hydrophobic residues such as leucine and phenylalanine. As expected, the P1 pyridinyl ring occupies the S1 pocket, which is known for its strict preference for glutamine. While most M pro inhibitors bear a pyrrolidinone glutamine mimetic at the P1 position, we determined that more hydrophobic residues can also bind to the S1 site, and that hydrogen bond formation with His163 is critical for inhibition 8 . In this instance, the pyridinyl ring of 23R is nearly superimposable with the same moiety from calpain inhibitor XII (Fig. 8c) However, the hydrophobic nature of the benzyl ring in 23R causes it to project towards the core near the S2 pocket, forcing Gln189 to rotate outwards (Fig. 8d) . This conformation is reinforced by pi-stacking interactions with the first phenyl of the biphenyl substituent. Notably, the binding pose of 23R features continuous intramolecular pi-stacking, where the phenyl is sandwiched by furan and benzyl groups, potentially contributing to its potent inhibition of M pro . Meanwhile, the S4 pocket remains largely unoccupied by 23R, leaving room for further improvement. In summary, the X-ray crystal structure of SARS-CoV-2 M pro in complex with 23R revealed two interesting structural features: 1) The P2 biphenyl is probably the largest substitution that can be accommodated in the S2 pocket, which is consistent with our design hypothesis.

2) The benzyl group from the terminal α -methylbenzyl fits in a previously unexplored binding pocket located in between the S2 and S4 pockets. As such, the benzyl group faces towards the core of the enzyme instead of solvent-exposed as seen with other existing M pro inhibitors. Although this is unexpected from the design perspective, this novel binding mode suggests that the new binding pocket in between S2 and S4 that can be explored for inhibitor design. 6XFN) . The highlights of this study include: 1) The overlaying X-ray crystal structures with multiple M pro inhibitors revealed the chemical space that can be explored for drug design. 2) All designed compounds were synthesized by the one-pot Ugi-4CR, which greatly facilitated the lead optimization. Indeed, we were able to improve the enzymatic inhibition potency by 35-fold from a focused library of 39 compounds. This is a significant advantage compared to covalent inhibitors such as GC376, which involves at least a five-step synthesis. 3) By introducing the chiral isocyanide, the diastereomer product can be conveniently separated by either silica gel column or reverse phase HPLC column, bypassing the need for an expensive chiral HPLC column. This greatly speeds up the co-crystallization. 4) The X-ray crystal structure of SARS-CoV-2 M pro in complex with 23R reveals a new binding pocket in between S2 and S4 sites that can be explored for drug design. Overall, using the expedited drug discovery approach, this study revealed a promising non-covalent M pro inhibitor 23R with a confirmed mechanism of action and potent cellular antiviral activity for further development. The expression and purification of SARS CoV-2 papain-like protease (PL pro ) was also described in our previous publications 7, 8, 18 .

The SARS-CoV-2 M pro FRET substrate Dabcyl-KTSAVLQ/SGFRKME(Edans) was synthesized as described before. 7 The SARS-CoV-2 PL pro FRET substrate Dabcyl-FTLRGG/APTKV(Edans) was synthesized by solid-phase synthesis through iterative cycles of coupling and deprotection using the previously optimized procedure. 20

Compound synthesis and characterization. Details for the synthesis procedure and characterization for compounds can be found in the supplementary information.

Native Mass Spectrometry. Prior to analysis, the protein was buffer exchanged into 0. Immediately before the viral inoculation, the tested compounds in a three-fold dilution concentration series were also added to the wells in triplicate. The infection proceeded for 48 h without the removal of the viruses or the compounds. The cells were then fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton-100, blocked with DMEM containing 10% FBS, and stained with a rabbit monoclonal antibody against SARS-CoV-2 NP (GeneTex, GTX635679) and an Alexa Fluor 488-conjugated goat anti-mouse secondary antibody (ThermoFisher Scientific). Hoechst 33342 was added in the final step to counterstain the nuclei.

Fluorescence images of approximately ten thousand cells were acquired per well with a 10x objective in a Cytation 5 (BioTek). The total number of cells, as indicated by the nuclei staining, and the fraction of the infected cells, as indicated by the NP staining, were quantified with the cellular analysis module of the Gen5 software (BioTek). X-ray diffraction data for the SARS-CoV-2 M pro structures were collected on the SBC 19-ID beamline at the Advanced Photon Source (APS) in Argonne, IL, and processed with the HKL3000 software suite. The CCP4 versions of MOLREP 24 was used for molecular replacement using a previously solved SARS-CoV-2 M pro structure, 6YB7. Structural refinement was performed using REFMAC5 25 and COOT 26 . The crystallographic statistics is shown in Table S1 .

The complex structure for SARS-CoV-2 M pro with 23R has been deposited in the Protein Data

Bank with the accession ID of 7KX5 (SARS-CoV-2 M Pro + Jun8-76-3A).

Baricitinib plus Remdesivir for Hospitalized Adults with Covid-19

COVID-19: Drug Targets and Potential Treatments

The SARS-CoV-2 main protease as drug target

SARS-CoV-2 Mpro inhibitors and activity-based probes for patient-sample imaging

Discovery of a Novel Inhibitor of Coronavirus 3CL Protease as a Clinical Candidate for the Potential Treatment of COVID-19. bioRxiv, 2020

GC-376, and calpain inhibitors II, XII inhibit SARS-CoV-2 viral replication by targeting the viral main protease

Structure and inhibition of the SARS-CoV-2 main protease reveals strategy for developing dual inhibitors against Mpro and cathepsin L

3C-like protease inhibitors block coronavirus replication in vitro and improve survival in MERS-CoV-infected mice

Efficacy of a 3C-like protease inhibitor in treating various forms of acquired feline infectious peritonitis

Reversal of the Progression of Fatal Coronavirus Infection in Cats by a Broad-Spectrum Coronavirus Protease Inhibitor

Discovery, synthesis, and structure-based optimization of a series of N-(tert-butyl)-2-(N-arylamido)-2-(pyridin-3-yl) acetamides (ML188) as potent noncovalent small molecule inhibitors of the severe acute respiratory syndrome coronavirus (SARS-CoV) 3CL protease

Discovery of N-(benzo[1,2,3]triazol-1-yl)-N-(benzyl)acetamido)phenyl) carboxamides as severe acute respiratory syndrome coronavirus (SARS-CoV) 3CLpro inhibitors: Identification of ML300 and noncovalent nanomolar inhibitors with an induced-fit binding

An automatic pipeline for the design of irreversible derivatives identifies a potent SARS-CoV-2 Mpro inhibitor. bioRxiv

Crystal structure of SARS-CoV-2 main protease provides a basis for design of improved alpha-ketoamide inhibitors

Structure of M(pro) from SARS-CoV-2 and discovery of its inhibitors

Crystal structure of SARS-CoV-2 main protease in complex with the natural product inhibitor shikonin illuminates a unique binding mode

Disulfiram, Carmofur, PX-12, Tideglusib, and Shikonin Are Nonspecific Promiscuous SARS-CoV-2 Main Protease Inhibitors

Evaluation of SARS-CoV-2 3C-like protease inhibitors using self-assembled monolayer desorption ionization mass spectrometry

Specific binding of adamantane drugs and direction of their polar amines in the pore of the influenza M2 transmembrane domain in lipid bilayers and dodecylphosphocholine micelles determined by NMR spectroscopy

Bayesian deconvolution of mass and ion mobility spectra: from binary interactions to polydisperse ensembles

Neutral red uptake assay for the estimation of cell viability/cytotoxicity

Pharmacological Characterization of the Spectrum of

Antiviral Activity and Genetic Barrier to Drug Resistance of M2-S31N Channel Blockers

MOLREP: an automated program for molecular replacement

REFMAC5 for the refinement of macromolecular crystal structures

Coot: model-building tools for molecular graphics

This research was partially supported by the National Institutes of Health (NIH) (Grants 

J. W., N. K., and C. M. are inventors of a patent claiming the use of compounds 23R and related compounds as potential SARS-CoV-2 antivirals.

Supplementary information accompanies this paper at