key: cord-0764502-yf59aqi2
authors: Xiong, Muya; Nie, Tianqing; Shao, Qiang; Li, Minjun; Su, Haixia; Xu, Yechun
title: In silico screening-based discovery of novel covalent inhibitors of the SARS-CoV-2 3CL protease
date: 2022-01-23
journal: Eur J Med Chem
DOI: 10.1016/j.ejmech.2022.114130
sha: 9b26b2107158c7a085c9a5a691346409a3f31f64
doc_id: 764502
cord_uid: yf59aqi2

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) 3CL protease (3CL(pro)) has been regarded as an extremely promising antiviral target for the treatment of coronavirus disease 2019 (COVID-19). Here, we carried out a virtual screening based on commercial compounds database to find novel covalent non-peptidomimetic inhibitors of this protease. It allowed us to identify 3 hit compounds with potential covalent binding modes, which were evaluated through an enzymatic activity assay of the SARS-CoV-2 3CL(pro). Moreover, an X-ray crystal structure of the SARS-CoV-2 3CL(pro) in complex with compound 8, the most potent hit with an IC(50) value of 8.50 μM, confirmed the covalent binding of the predicted warhead to the catalytic residue C145, as well as portrayed the interactions of the compound with S1’ and S2 subsites at the ligand binding pocket. Overall, the present work not merely provided an experiment-validated covalent hit targeting the SARS-CoV-2 3CL(pro), but also displayed a prime example to seeking new covalent small molecules by a feasible and effective computational approach.

Since coronavirus disease 2019 (COVID-19) caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) became prevalent in the world [1, 2] , more than 249 million confirmed cases including 5 million deaths have been reported up to 8 th November, 2021 (https://covid19.who.int). Although several vaccines targeting the spike protein, two inhibitors of the SARS-CoV-2 RNA-dependent RNA polymerase (RdRp), remdesivir and molnupiravir, together with an inhibitor of the SARS-CoV-2 3C-like protease (3CL pro ), nirmatrelvir [PF-07321332], have been approved to combat SARS-CoV-2 [3] [4] [5] [6] , it still needs to find much more therapeutic interventions for this global pandemic. A chymotrypsin-like protease called 3CL pro , also referred to as the main protease, is a prominent protease which cleaves polyproteins to generate mature nonstructural proteins involved in the virus replication and transcription of coronaviruses [7, 8] . Besides, the rigorous specificity for recognizing the P1-Gln substrate residue at the cleavage site endows the high conservation of the ligand binding J o u r n a l P r e -p r o o f site of 3CL pro among known coronaviruses [9] [10] [11] . Hence, 3CL pro has been regarded as a pivotal therapeutic target for treating COVID-19 and other coronavirus-caused diseases [12] , and accordingly, the development of 3CL pro inhibitors has attracted much attention from medicinal chemists and pharmaceutical industry [13] [14] [15] [16] [17] .

Currently, the majority of available 3CL pro inhibitors are peptidomimetic compounds [18] [19] [20] , which were designed by the addition of a covalent warhead to a substrate mimic. The known SARS-CoV-2 3CL pro inhibitors like N3 [9, 13] , 11a [16] , 13b [14] , PF-00835231 [21] , PF-07321332 [6] , and the TGEV 3CL pro inhibitor such as Cbz-VNSTLQ-CMK [22] , contain a warhead of Michael acceptor, aldehyde, αketoamide, hydroxymethylketone, nitrile, and chloromethyl ketone, respectively. In contrast, non-peptidomimetic inhibitors have been rather derived from high-throughput screening/virtual screening of repurposing drugs/natural products/compound database [23] [24] [25] . Given that the presence of the covalent warhead endows the superiority in prolonged residence time [26] and it is a big challenge to improve the oral bioavailability of peptidomimetic inhibitors, non-peptidomimetic small-molecule inhibitors carrying covalent warheads may offer an opportunity to facilitate the discovery of more potent and drug-like 3CL pro inhibitors.

Several SARS-CoV-2 3CL pro non-peptidomimetic covalent inhibitors like ebselen [13] , PX-12 [13] , carmofur [23] , myricetin [25] , and ester derivatives [27] have been identified mostly by the high-throughput screening. However, there are very few studies that such inhibitors were uncovered or designed with the aid of computational approaches. Recently, Daniel, et al. [28] successfully employed a computational pipeline, covalentizer, to design a low micromolar covalent SARS-CoV-2 3CL pro inhibitor by adding an acrylamide warhead to a non-covalent inhibitor, ML188.

Similarly, Stille et al. [29] utilized the docking program FITTED to convert a noncovalent inhibitor of the SARS-CoV 3CL pro (X77) to a sub-micromolar covalent inhibitor by replacing the imidazole group with a covalent warhead. In addition, a covalent virtual screening on an in-house database of small pseudopeptides with Michael acceptors yielded two SARS-CoV-2 3CL pro inhibitors, albeit with the relatively low potency [30] .

In this work, we collected covalent warheads observed among the known peptidomimetic inhibitors of the SARS-CoV-2 3CL pro and used them as reference to pick up the potential covalent binders from the list resulting from the virtual screening on ChemDiv database. A covalent molecular docking was further performed on these J o u r n a l P r e -p r o o f selected potential covalent binders. Finally, we purchased eight compounds for testing their enzymatic activities against the SARS-CoV-2 3CL pro , and yielded three novelscaffold hits with different warheads. The X-ray crystal structure of the SARS-CoV-2 3CL pro in complex with the most effective compound (8) revealed the covalent binding mode, and the intrinsic reactivity of its warhead was explored through a glutathione (GSH) assay, providing a basis for further optimization. Accordingly, our study demonstrated that the strategy of a virtual screening followed by a covalent molecular docking is able to quickly identify the novel covalent inhibitors of the SARS-CoV-2 3CL pro .

Our protocol for in silico screening consisted of two stages: a large-scale noncovalent virtual screening of compounds from ChemDiv database, which was followed by a covalent docking of the compounds containing the warheads shown in Figure 1 .

We used tools implemented in the Schrödinger 2015 suite to carry out the virtual screening with the crystal structure of SARS-CoV-2 3CL pro in complex with myricetin (PDB ID: 7B3E) as the receptor structure. The receptor was prepared using the Protein Preparation Wizard [31] to add hydrogen atoms and the missing residue side chains. The covalent bond formed between the catalytic C145 and myricetin was broken for proceeding with non-covalent virtual screening in the first stage. The overall structure was refined using OPLS3 forced field [32] with harmonic restraints on heavy atoms. The receptor grid was centered on the centroid of myricetin and generated using the Receptor Grid Generation to define the binding site for ligand docking. Subsequently, a virtual screening was performed using Virtual Screening Workflow ( Figure 1 ). The three-dimensional (3D) structures of ∼1,500,000 compounds from the database were yielded with LigPrep. The overall workflow included filtering the prepared structures of compounds with Propfilter and docking them to the receptor grid using Glide. In the filtering step, we filtered out compounds that are not compatible with Lipinski's Rule of 5. In the docking step, we carried out standard precision (SP) docking with default settings. The top 5000 docked compounds were then reserved for the extra precision (XP) docking and the 3000 highest-ranking compounds out of them were retained for covalent docking in the second stage.

Using FAF-Drug4 Server [33] , 173 compounds containing potential covalent warheads were first selected from 3000 compounds which were collected from the noncovalent virtual screening. Furthermore, 160 compounds containing five different types of warheads including chloromethyl ketone, Michael acceptor, aldehyde, nitrile, and alpha-ketoamide, which are widely applied in the reported 3CL pro inhibitors [10, 18, 20] and also easy to be synthesized, were selected, and their 3D structures were produced J o u r n a l P r e -p r o o f using LigPrep.

The structure of SARS-CoV-2 3CL pro -myricetin complex as mentioned above was used as the receptor structure, and the protease-inhibitor covalent bond was maintained for the following covalent docking using Covalent Docking [34] . The catalytic C145 was defined as reactive residue and the centroid of myricetin was used as the box center.

The reaction type was determined by the reactive group of the compound. And the covalent docking was performed with a pose prediction mode. The top-ranking compounds were visually checked to find the correct reactive sites as well as binding modes, particularly, to meet the requirement for interacting with the residues in the crucial S1 (F140/N142/G143/H163/E166) or S2 (H41/M49/Q189) subsites as suggested by the structure-activity relationship summarized in our previous review [20] .

In the end, we purchased 8 compounds with five types of covalent warheads and different scaffolds from TargetMol for an inhibitory activity test. PF-07321332, a positive control to our enzymatic inhibition assay, was purchased from MedChemExpress.

The cDNA of SARS-CoV-2 3CL pro (Gen-Bank: MN908947.3) with a N-terminal SUMO tag was cloned into the pET-15b vector. The plasmid was then transformed into BL21 (DE3) cells for protein expression. The expressed protein was purified by a Ni-NTA column (GE Healthcare) and cleaved by the SUMO specific peptidase 2 (SENP2) to remove the SUMO tag. The resulting protein sample was further purified by Q-Sepharose followed by a size-exclusion chromatography (GE Healthcare). The eluted protein samples were stored in a solution (10 mM Tris, pH 7.5) for the enzymatic inhibition assay and protein crystallization.

A fluorescence resonance energy transfer (FRET) protease assay was applied to measure the inhibitory activity of compounds against the SARS-CoV-2 3CL pro [24, 25] . 

The purified SARS-CoV-2 3CL pro protein was concentrated to 9 mg/mL for crystallization. To obtain complex structures, the SARS-CoV-2 3CL pro protein was incubated with 8 mM compound 8 for 1 h before crystallization condition screening.

Crystals of the complex was obtained under the condition of 10-25% PEG6000, 100 mM MES, pH 5.75-6.25, and 3% DMSO. Crystals were flash frozen in liquid nitrogen in the presence of the reservoir solution supplemented with 20% glycerol. X-ray diffraction data were collected at beamline BL18U1 at the Shanghai Synchrotron Radiation Facility [35] . Bluice was used to collect X-ray diffraction data. The data were processed with HKL3000 software packages [36] . The complex structures were solved by molecular replacement using the program PHASER [37] with a search model of PDB code 6M2N [24] . The model was built using Coot [38] and refined with the program PHENIX [39] . The refined structure was deposited to Protein Data Bank with the accession code listed in Table 2 . The complete statistics as well as the quality of the solved structure are also shown in Table 2 .

The reaction rates of the compounds with GSH were measured with the previous reported protocol [25] . The compounds 2, 3 and 8 at a final concentration of 400 μM were incubated with 10 μM GSH in potassium phosphate buffer, respectively. 1Hpyrrole-2,5-dione (CPM, ThermoFisher) at a concentration of 50 μM was added to the reaction system at definite time (2.5, 5, 10, 20, 40, 60, and 80 min) to quantify the remaining GSH. After that, the fluorescence signal at 384 nm (excitation)/470 nm (emission) was immediately measured using a Bio-Tek Synergy4 plate reader. Ln (the percentage of the remaining compounds) was plotted against incubation time to generate the half-life time of the compound reacting with GSH.

J o u r n a l P r e -p r o o f As described in Materials and Methods, 8 compounds resulting from the noncovalent virtual screening and covalent docking were subject to the experimental validation using a FRET protease assay. In the initial test at a concentration of 50 µM, 3 of 8 compounds (2, 3 and 8) displayed over 50% inhibition towards the SARS-CoV-2 3CL pro (Figure 2A and Table 1 

We then sought to determine crystal structures of the SARS-CoV-2 3CL pro in complex with 3 hit compounds in order to elucidate the precise protease-ligand binding mode. What's more, it also would be revealed directly whether these hits are covalent inhibitors or not. As a result, the crystal structure of the SARS-CoV-2 3CL pro bound with compound 8, the best inhibitor among three compounds (IC50 = 8.50 µM), was determined. As expected, the methylene group of the chloromethyl ketone moiety was attacked by the catalytic C145 to form a C-S covalent bond, supported by continuous electron density map between the inhibitor and the residue ( Figure 3A) . J o u r n a l P r e -p r o o f The protease is rendered in grey cartoon, compound 8 is shown in orange sticks, and the surrounding residues are displayed with light-purple sticks. H-bonding, π-π stacking and halogen-bonding interactions are represented by black, blue and purple dashed lines, respectively. A molecular surface representation of compound 8 interacting with the S1ʹ-S4 subsites of the protease in a (C) top view and (D) front view. The subsites are colored purple (S1ʹ), pink (S1), yellow (S2), and green (S4). The catalytic dyad of the protease (C145 and H41) is shown in sticks.

The substrate binding site of the SARS-CoV-2 3CL pro is normally divided into S1', S1, S2, and S4 subsites [10, 15] . As shown in Figure 3C , compound 8 mainly occupied the S1' and S2 subsites. Particularly, the highly plastic S2 subsite, featured with the flexible Q189 in the SARS-CoV 3CL pro [40] , perfectly accommodated the m-Cl substituted benzene ring of 8, making it to be sandwiched between H41 and Q189

( Figure 3D ). It is well acknowledged that the S1 subsite of 3CL pro specifically recognizes the P1-Gln of the substrate so that almost all the peptidomimetic 3CL pro inhibitors used mimics of glutamine to interact with the S1 subsite [41, 42] . Such a strategy could also be utilized in the optimization of compound 8 to pursuit the higher potency.

Since all attempts failed to determine crystal structures of the protease bound with Figure 4B ).

Accordingly, both compounds 2 and 3 mainly occupied the S1' and S1 subsites ( Figure 4C, D) , which is distinct to the binding pose of compound 8 described above.

Moreover, the 5-bromo-benzoic acid group of compound 2 reached into the S2 and S4

subsites, while the benzo[d] [1, 3] dioxole group of compound 3 was only located at the border of S1', S2 and S4 without protruding into any of these subsites ( Figure 4C, D) . interacting with S1ʹ-S4 subsites of 3CL pro from a top view. The subsites are colored purple (S1ʹ), pink (S1), yellow (S2), and green (S4). The catalytic dyad is shown in sticks.

Generally speaking, the reactivity of covalent inhibitors with the targeted cysteine is crucial for covalent drug development, considering that the hyper-reactivity of the warheads may trigger potential side effects, e.g., drug-induced toxicity or immunogenicity [43, 44] . Consequently, the reactivity of three hit compounds was measured by a GSH assay which is usually used to estimate the reactivity of cysteinetargeted warheads [45] . The resulting half-lives (t1/2) of compounds 3 and 8 were 103

and 90 min, respectively, while that of compound 2 was less than 2.5 min ( Figure 5 ).

Referring to the fact that t1/2s of covalent kinase inhibitors in clinics are often in the range of 30 to 512 min [46] , 

In the present work, the strategy of a non-covalent virtual screening combined with a covalent docking was utilized to seek out novel covalent non-peptidomimetic inhibitors of the SARS-CoV-2 3CL pro . Based on the results from such a combined screening, we purchased eight compounds with five types of covalent warheads which were previously reported in covalent peptidomimetic inhibitors of 3CL pro . Following the validation of the enzyme inhibition, we succeeded in unveiling three hits with IC50 values of 19.09, 36.07 and 8.50 µM, respectively. Coincidentally, these active compounds, namely compounds 2, 3 and 8, were armed with three divergent covalent warheads including nitrile, Michael acceptor, and chloromethyl ketone. Remarkably, the crystal structure determination of the 3CL pro -8 complex depicted that compound 8 covalently binds to C145 and meanwhile non-covalently interacts with residues of the S1' and S2 subsites. In contrast, compounds 2 and 3 were predicted to mainly interact with the S1' and S1 subsites. Furthermore, the reactivity of 3 hits evaluated by the GSH assay demonstrated that the warheads of compounds 3 and 8 have the moderate reactivity to GSH, rendering them to be further developed as potent covalent inhibitors of the protease. Overall, the identified compound 8 provided a good starting point for the discovery of novel covalent non-peptidomimetic inhibitors of the SARS-CoV-2 3CL pro , and the covalent docking offers an efficient and feasible strategy for discovery of novel covalent binders.

The authors declare no competing interests.

Immunity, endothelial injury and complement-induced coagulopathy in COVID-19

A novel coronavirus outbreak of global health concern

Remdesivir and chloroquine effectively inhibit the recently emerged novel coronavirus (2019-nCoV) in vitro

Remdesivir for the treatment of COVID-19-final report

Molnupiravir for Oral Treatment of Covid-19 in Nonhospitalized Patients

An oral SARS-CoV-2 M pro inhibitor clinical candidate for the treatment of COVID-19

Structure and inhibition of the SARS-CoV-2 main protease reveal strategy for developing dual inhibitors against M pro and cathepsin L

Discovery of ketone-based covalent inhibitors of coronavirus 3CL proteases for the potential therapeutic treatment of COVID-19

Design of wide-spectrum inhibitors targeting coronavirus main proteases

An overview of Severe Acute Respiratory Syndrome-Coronavirus (SARS-CoV) 3CL protease inhibitors: Peptidomimetics and small molecule chemotherapy

The newly emerged SARS-like coronavirus HCoV-EMC also has an "Achilles' heel": current effective inhibitor targeting a 3C-like protease

Research and development on therapeutic agents and vaccines for COVID-19 and related human coronavirus diseases

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

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

SARS-CoV-2 M pro inhibitors with antiviral activity in a transgenic mouse model

Structure-based design of antiviral drug candidates targeting the SARS-CoV-2 main protease

X-ray screening identifies active site and allosteric inhibitors of SARS-CoV-2 main protease

Recent discovery and development of inhibitors targeting coronaviruses

The development of Coronavirus 3C-Like protease (3CL(pro)) inhibitors from

What coronavirus 3C-like protease tells us: From structure, substrate selectivity, to inhibitor design

Discovery of a novel inhibitor of coronavirus 3CL protease for the potential treatment of COVID-19

Coronavirus main proteinase (3CLpro) structure: basis for design of anti-SARS drugs

Structural basis for the inhibition of SARS-CoV-2 main protease by antineoplastic drug carmofur

Anti-SARS-CoV-2 activities in vitro of Shuanghuanglian preparations and bioactive ingredients

Identification of pyrogallol as a warhead in design of covalent inhibitors for the SARS-CoV-2 3CL protease

Prolonged and tunable residence time using reversible covalent kinase inhibitors

Indole chloropyridinyl esterderived SARS-CoV-2 3CLpro inhibitors: Enzyme inhibition, antiviral efficacy, structure-activity relationship, and X-ray structural studies

An automatic pipeline for the design of irreversible derivatives identifies a potent SARS-CoV-2 M pro inhibitor

Moitessier, Design, synthesis and in vitro evaluation of novel SARS-CoV-2 3CLpro covalent inhibitors

Lead discovery of SARS-CoV-2 main protease inhibitors through covalent docking-based virtual screening

Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments

OPLS3: A force field providing broad coverage of drug-like small molecules and proteins

FAF-Drugs4: free ADME-tox filtering computations for chemical biology and early stages drug discovery

Docking covalent inhibitors: A parameter free approach to pose prediction and scoring

Upgrade of macromolecular crystallography beamline BL17U1 at SSRF

HKL-3000: the integration of data reduction and structure solution -from diffraction images to an initial model in minutes

Phaser crystallographic software

Coot: model-building tools for molecular graphics

PHENIX: building new software for automated crystallographic structure determination

Alpha-Ketoamides as broad-spectrum inhibitors of coronavirus and enterovirus replication: Structurebased design, synthesis, and activity assessment

Structure-based design, synthesis, and evaluation of peptidemimetic SARS 3CL protease inhibitors

Comprehensive insights into the catalytic mechanism of Middle East Respiratory Syndrome 3C-Like protease and Severe Acute Respiratory Syndrome 3C-Like potease

A road map for prioritizing warheads for cysteine targeting covalent inhibitors

Covalent inhibitors in drug discovery: from accidental discoveries to avoided liabilities and designed therapies

Comparative reactivity analysis of small-molecule thiol surrogates

-yl)-2-(1H-indol-1-yl)acetamides as covalent inhibitors of KRAS(G12C)

We thank the staff from BL18U1 beamline of National Facility for Protein Science in