Discovery of novel inhibitors of SARS-CoV-2 main protease

Abstract

Corona Virus Disease 2019 (COVID-19), referred to as ‘New Coronary Pneumonia’, is a type of acute infectious disease caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) infection. M^pro is one of the main targets for treating COVID-19. The current research on M^pro mainly focuses on the repurposing of old drugs, and there are only a few novel ligands that inhibit M^pro. In this research, we used computational free energy calculation to screen a compound library against M^pro, and discovered four novel compounds with the two best compounds (AG-690/13507628 and AG-690/13507724) having experimental measured IC₅₀ of just under 3 μM and low cell toxicity. Detailed decomposition of the interactions between the inhibitors and M^pro reveals key interacting residues and interactions that determine the activity. The results from this study should provide a basis for further development of anti-SARS-CoV-2 drugs.

Communicated by Ramaswamy H. Sarma

Introduction

Coronavirus Disease (COVID-19), which is referred as ‘New Coronary Pneumonia’, is caused by the infection of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2). The main symptoms of COVID-19 include respiratory symptoms, fever, cough, and breathing difficulties (Struyf et al., 2020). In more severe cases, the infection can lead to pneumonia, severe acute respiratory syndrome, kidney failure, and even death (Wang et al., 2020). Since the outbreak of the pandemic in later 2019, the coronavirus has been rapidly discovered in the world because of its own pathogenic characteristic and infectious characteristic. The number of confirmed infections has surpassed that of SARS, and the number of deaths has continued to increase, causing serious social harm and economic losses. The World Health Organization announced that the outbreak of New Coronary Pneumonia has constituted a ‘Public Health Emergency of International Concern’ (“Coronavirus disease,” 2020). Therefore, the development of therapeutic drugs against the SARS-CoV-2 is a very urgent task.

The genome of SARS-CoV-2, which likes most of the Coronavirus genome, encodes two large polyproteins, namely PP1A and PP1AB (Thiel et al., 2003). These two polyproteins are cleaved by two proteases M^pro (3 C-like protease or 3CL^pro) and PL^pro (papain-like protease) which are encoded by the ORF 1a/b (Hilgenfeld, 2014). The main protease M^pro is crucial for virus replication and controlling of host cell response, which makes it important for the propagation of the virus. Therefore, it is considered as the one of the key targets in the development of anti-SARS-CoV-2 drugs.

M^pro is a dimer with monomer contains two domains: the N-terminal catalytic domain and the C-terminal domain (Lee et al., 2005). The sequences of M^pro in SARS-CoV and SARS-CoV-2 have a sequence identity of 96% (Gil et al., 2020). Around the active site, the conserved binding pocket of M^pro is composed of four subsites (S1’, S1, S2, and S4) well accommodating the substrate (Figure 1) (Xue et al., 2008). Moreover, the high-resolution crystal structure of SARS-CoV-2 M^pro in complex with the inhibitor N3 has been solved (Protein Data Bank ID: 7BQY) (Jin et al., 2020), which provides an excellent basis for structure-based drug design.

Figure 1.

The structure of SARS-CoV-2 M^pro. The coordinates are from PDB entry 7BQY, and the N3 inhibitor is shown in stick. The substrate-binding pocket is shown in the inset with subsites labeled.

Recently, a number of inhibitors against M^pro have been discovered. For example, Jin et al. reported that the IC₅₀ of Ebselen, an anti-inflammatory drug (Jin et al., 2020), was 0.67 μM. In the study of Li et al., several old drugs like Dipyridamole, Candesarta, and Cilexetil were found to have a inhibitory effect in vitro (Li et al., 2020). Moreover, various inhibitors against SARS-CoV-1 M^pro and other antiviral drugs including Boceprevir, Nariaprevir, GC-376, MG-132, Calpain inhibitor II, and Calpain inhibitor XII have been proven effective against the M^pro of SARS-CoV2 in vitro (Ma et al., 2020). A number of novel inhibitors of M^pro have also been discovered by research groups. Dai et al. discovered several novel polypeptide analogues with IC₅₀ of approximately 0.05 μM (Wenhao et al., 2020). The alpha-ketoamide inhibitors such as 13 b have good inhibitory activity with IC₅₀ of 0.67 μM (Linlin Zhang et al., 2020). Small-molecule compounds such as N-substituted isatin have shown enzymatic activity at sub-micromolar level (Liu et al., 2020). An indole moiety inhibits for M^pro could block the infectivity of SARS-CoV-2 with EC₅₀ values of 15 ± 4 and 4.2 ± 0.7 μM for GRL-1720 and 5 h (Hattori et al., 2021). Recently, Qiao et al. discovered six compounds of the M^pro inhibitors with nanomolar or low micromolar EC50 values, including MI-09 (0.86 μM), MI-12 (0.53 μM), MI-14 (0.66 μM), MI-28 (0.67 μM), MI-30 (0.54 μM), and MI-31 (0.83 μM) (Qiao et al., 2021). Zhang et al. obtained a series of non-covalent inhibitors by modifying Perampanel, and the IC₅₀ of best molecule reached 18 nM (Zhang et al., 2021). Despite these exciting progresses, discovering novel M^pro inhibitors to enrich our arsenal against SARS-CoV-2 is still in urgent need.

In this study, we used virtual screening to identify M^pro inhibitors from the Specs compound library. Four active compounds with a novel scaffold and micromolar IC₅₀ values were discovered. Furthermore, these compounds have multiple modifiable chemical sites, low cytotoxicity and potentials for further optimization.

Methods

Virtual screening

The virtual screening was performed using Glide (Friesner et al., 2004). The structure of SARS-CoV-2 M^pro (PDB ID:7BQY) (Jin et al., 2020) was used to generate the receptor grid for docking simulations. The center of the grid was determined using the inhibitor N3 in the structure with a grid box of 30 Å. The Van der Waals radius scaling factor was 1.0 and partial charge cutoff was 0.25. We used the Specs library (Delft, Netherlands: http://www.specs.net) the ligand library. The ligands were prepared with LigPrep (Friesner et al., 2004) at target pH 7.2 ± 0.2 with the OPLS3e force field (Harder et al., 2016).

Molecular dynamics simulations

We used the gaff force field to obtain parameters for the compounds (Wang et al., 2004). The charges for the all the ligands were calculated using the AM1-BCC method in antechamber of Amber18 (Salomon-Ferrer et al., 2013). Protonation of the protein-ligand complex was performed at pH 7.4. Molecular dynamics (MD) simulations were carried out using Amber18 with the ff14SB force field (Maier et al., 2015). We solvated the protein-ligand complex in a cubic box using TIP3P water, which extends 10.0 Å away from the solute. Cut-off of non-bonded interactions was set to 10 Å. We added counter-ions to neutralize the systems. Each system was minimized for 10,000 steps and then heated from 0 to 300 K in 300 ps with Langevin dynamics. Berendsen barostat was used to control the pressure at 1.0 atm. Production runs were 10-ns long in an NPT ensemble with a time step of 2 fs. The trajectories were recorded every 1 ps. We used the SHAKE algorithm to constrain the bonds involving hydrogen atoms (Andersen, 1983).

Binding free energy calculation

We used the alanine-scanning in combination with interaction entropy (ASIE) method to calculate the binding free energy (Yan et al., 2017). In this approach, it is assumed that the contribution of the mutated alanine to the binding free energy is negligible. Therefore, the contribution of a specific residue (Zhou et al., 2018) to the total binding free energy is given by the energy difference when the wildtype residue X is mutated to A (alanine):

$$\Delta \Delta {\mathrm{G}}_{\text{bind}}^{\mathrm{x}\to a}=\Delta {\mathrm{G}}_{\text{bind}}^a-\Delta {\mathrm{G}}_{\text{bind}}^x=\Delta \Delta {\mathrm{G}}_{\text{gas}}^{\mathrm{x}\to a}+\Delta \Delta {\mathrm{G}}_{\text{sol}}^{\mathrm{x}\to a}$$ (1)

The gas-phase binding free energy and the solvation component are evaluated by the following scheme:

$$\Delta \Delta {\mathrm{G}}_{\text{gas}}^{\mathrm{x}\to a}=\Delta {\mathrm{G}}_{\text{gas}}^a-\Delta {\mathrm{G}}_{\text{gas}}^{\mathrm{x}}$$ (2)

$$\Delta \Delta {\mathrm{G}}_{\text{sol}}^{\mathrm{x}\to a}=\Delta {\mathrm{G}}_{\text{sol}}^a-\Delta {\mathrm{G}}_{\text{sol}}^{\mathrm{x}}$$ (3)

The gas-phase component of the binding free energy is $\Delta {\mathrm{G}}_{\text{gas}}^a$ and $\Delta {\mathrm{G}}_{\text{gas}}^{\mathrm{x}},$ which represent the energy between the ligand L and alanine and the wildtype residue X. The enthalpy and entropy of gas-phase energy were calculated with the standard molecular mechanics and the IE method (Ben-Shalom et al., 2017; Duan et al., 2016; Qiu et al., 2018; Song et al., 2018), respectively

$$\Delta {\mathrm{G}}_{\text{gas}}^{\mathrm{x}}=〈{\mathrm{E}}_{\text{int}}^{\mathrm{x}}〉-\mathrm{T}\Delta {\mathrm{S}}_{\mathrm{int}}^x=〈{\mathrm{E}}_{\text{int}}^{\mathrm{x}}〉-K\text{Tln}〈{\mathrm{e}}^{\beta \Delta {\mathrm{E}}_{\text{int}}^{\mathrm{x}}}〉$$ (4)

$$\Delta {\mathrm{G}}_{\text{gas}}^{\mathrm{a}}=〈{\mathrm{E}}_{\text{int}}^{\mathrm{a}}〉-K\text{Tln}〈{\mathrm{e}}^{\beta \Delta {\mathrm{E}}_{\text{int}}^{\mathrm{a}}}〉$$ (5)

where, ${\mathrm{E}}_{\text{int}}^{\mathrm{a}}$ and ${\mathrm{E}}_{\text{int}}^{\mathrm{x}}$ were the van der Waals and electrostatic interaction energies between the ligand and residue X and A, respectively. The exponential argument $\beta$ denotes $\frac{1}{\text{KT}},$ and $\Delta {\mathrm{E}}_{\mathrm{int}}^{\mathrm{x}/\mathrm{a}}$ is the deviation from average. We evaluated the exponential average ${\mathrm{E}}_{\text{int}}^{\mathrm{x}/\mathrm{a}}$ by discrete time averaging,

$$〈{\mathrm{e}}^{\beta \Delta {\mathrm{E}}_{\text{int}}^{\mathrm{x}}}〉=\frac{1}{\mathrm{N}}\sum _{\mathrm{i}=1}^{\mathrm{N}}{\mathrm{e}}^{\beta \Delta {\mathrm{E}}_{\text{int}}^{\mathrm{x}}({\mathrm{t}}_{\mathrm{i}})}$$ (6)

where N is the number of MD snapshots. Finally, Eq. (2) becomes

$$\Delta \Delta {\mathrm{G}}_{\text{gas}}^{\mathrm{x}\to \mathrm{a}}=\Delta \Delta E_{\text{gas}}^{\mathrm{x}\to \mathrm{a}}-T\Delta \Delta S_{\text{gas}}^{\mathrm{x}\to \mathrm{a}}=〈{\mathrm{E}}_{\mathrm{int}}^{\mathrm{a}}〉-〈{\mathrm{E}}_{\mathrm{int}}^{\mathrm{x}}〉+KT\left[\text{ln}\hspace{0.17em}〈{\mathrm{e}}^{\mathrm{\beta }\Delta {\mathrm{E}}_{\mathrm{int}}^{\mathrm{a}}}〉-\hspace{0.17em}\text{ln}\hspace{0.17em}〈{\mathrm{e}}^{\mathrm{\beta }\Delta {\mathrm{E}}_{\mathrm{int}}^{\mathrm{a}}}〉\right]$$ (7)

The MM/GBSA method (igb = 8 in Amber18) (Genheden & Ryde, 2015; Massova & Kollman, 2000; Onufriev et al., 2004) is used to calculate the solvation free energy by

$$\Delta G_{\mathit{sol}}=\Delta G_{gb}+\Delta G_{np}$$ (8)

where $G_{gb}$ and $G_{np}$ are the electrostatic solvation free energy and the nonpolar solvation free energy, respectively. $G_{np}$ is given by an empirical solvent-accessible surface area (SASA) formula:

$$G_{np}=\gamma \mathit{SASA}+\beta$$ (9)

The γ and β values we used here are the standard values of 0.00542 kcal/(mol·Å²) and 0.92 kcal/mol. In our ASIE calculation, different dielectric constant of 1, 3, and 5 are used for nonpolar, polar, and charged residues (Hou et al., 2011; Petukh et al., 2015).

The total protein-ligand binding free energy can be approximated by the summation (Liu et al., 2018; Zhou et al., 2018),

$$\Delta {\mathrm{G}}_{\text{bind}}=-\sum _{\mathrm{x}}\Delta \Delta {\mathrm{G}}_{\text{bind}}^{\mathrm{x}\to \mathrm{a}}$$ (10)

where the summation is over residues that are within 5 Å of the ligand. A total of 100 snapshots from the last 5 ns of the MD trajectory were used to calculate the binding energy with the ASIE method.

Activity of the protein

The SARS-CoV-2 M^pro protein was purchased from Kactus Biosystems Co, Ltd (Figure S1). PBS-EDTA (4 × PE, pH 7.4, LEAGEN) with 4 mM DTT (TCI, CAS RN 3483-12-3) was diluted as the buffer solution. We used a FRET-based M^pro substrate of SARS-CoV-2 Dabcyl-KTSAVLQ/SGFRKME (Edans) (Go Top Peptide Biotech Co, Ltd). According to the preliminary experiment read by an imaging reader (BioTek, SYNERGY4), we set filters for excitation at 340 nm and emission at 535 nm. We set up six concentrations of the substrate at 100 μM, 50 μM, 25 μM, 12.5 μM, 6.25 μM, and 3.125 μM to measure the activity of the protein. The reaction progress was monitored for 30 min. In the first 20 min, the reaction kinetics were monitored every 90 s, and the data was used to calculate the initial velocity via linear regression in Prism 8 with the classic Michaelis–Menten equation.

Enzymatic assays

For the screening of inhibitors and IC₅₀ measurements, 100 nM M^pro was incubated with the ligands at 30 °C for 60 min in reaction buffer, then the reaction was initiated by adding 5 µM FRET substrate and monitored for 1 h. The IC₅₀ value was calculated by plotting the initial velocity against various concentrations of ligands using a dose-response curve in the Prism 8 software.

The compound concentration was 500 μM, 100 μM, 10 μM, 1 μM, 0.1 μM, and 0.01 μM for the measurement the IC₅₀ values. The data from imaging reader was calculated with the equation below:

$$\text{Reaction}\mathrm{ }\text{Activity}\mathrm{(}\%\mathrm{)}={\mathrm{(}\text{RLU}}_{\text{compound}}-{\text{RLU}}_{0\%\text{control}})/{\mathrm{(}\text{RLU}}_{\text{DMSO}-\text{control}}-{\text{RLU}}_{0\%\text{control}})\times 100\%$$

where ${\text{RLU}}_{\text{compound}}$ stands for protease, drug, buffer and substrate; ${\text{RLU}}_{0\%\text{control}}$ stands for buffer and substrate; ${\text{RLU}}_{\text{DMSO}-\text{control}}$ stands for protease, DMSO, buffer and substrate.

Cytotoxicity test

Human Umbilical Vein Endothelial Cells (HUVEC) were seeded into a 96-well plate. After the cells adhere to the wall, the compounds were diluted to 100 μM and 500 μM for administration. Each concentration was tested with three replicates. CCK-8 reagent was added 72 h after the administration, and incubated for 2 h. Then the plate was read with an imaging reader (BioTek, SYNERGY4). The GraphPad Prism 8 software was used to calculate the cell activity inhibition plot for normal cells.

Results

Virtual screening

We used the complex structure of SARS-CoV-2 M^pro and the N3 inhibitor (PDB ID: 7BQY) (Jin et al., 2020) for in silico screening against the Specs library with Glide. The top 400 compounds from docking were selected for binding free energy calculation using the ASIE method. The top 200 compounds from ASIE were manually inspected, and 20 compounds were selected based on the calculated binding free energy from ASIE, 5 compounds were selected according to Glide XP score, and 5 were selected based on experience of medicinal chemistry. At last, 30 compounds were purchased from Specs (Table 1).

Table 1.

The Glide XP score of the 30 purchased compounds.

Specs ID	Glide XP score	Specs ID	Glide XP score
AE-562/12222311	–9.726	AK-918/42028521	–8.343
AE-848/07789006	–8.128	AK-918/42293923	–8.033
AF-399/34897016	–8.058	AL-398/12677067	–8.493
AG-205/12085061	–8.663	AO-022/43235300	–8.865
AG-650/41069129	–8.358	AO-081/15245021	–8.427
AG-670/40968109	–8.042	AO-081/41756050	–8.725
AG-690/08355055	–8.032	AO-554/14700005	–8.525
AG-690/11763097	–9.087	AP-064/41684917	–8.277
AG-690/13507628	–8.457	AP-064/41684959	–8.029
AG-690/40753764	–8.037	AP-064/42646016	–8.875
AG-690/40753996	–8.176	AP-124/40904362	–8.051
AH-487/15274490	–8.710	AP-970/13521174	–8.685
AJ-030/14523202	–8.317	AP-970/42898518	–8.898
AJ-292/13489193	–8.015	AR-360/42760471	–8.262
AK-778/43464919	–8.494	AT-057/43469634	–8.515

Biological activity of the compounds

We first tested the activity of the wildtype M^pro, and found that its K_m (119.5 μM) and V_max (55.61 nM/s) are close to the values reported in a previous study (Li et al., 2020) (Figure S2). The activity of the 30 compounds were then tested at two concentrations of 10 μM (Figure S3) and 100 μM. At 10 μM, one of the compounds, AG-690/13507628, was found to inhibit the protein activity by 70%, and its IC₅₀ value was determined to be 9.457 μM (Figure S4a). The IC₅₀ of the positive reference Ebselen was 0.786 μM in our assay (Figure S4b).

Figure 2.

The IC₅₀ of the four active compounds (95% Confidence Interval). Each experiment was repeated three times.

Figure 3.

Compound AG-690/13507628, AG-690/13508164, and AG-690/13507724 have little cytotoxicity on HUVEC cells at 100 µM. HUVEC cells were treated with the indicated compounds for 72 h and their cytotoxicity was determined by CCK8 assay.

Figure 4.

The free energy contributions of the dominant residues of M^pro in binding to four compounds.

We further purchased 30 compounds that share a similar skeleton with AG-690/13507628 in Specs and performed additional screening at concentrations of 10 μM (Figure S5) and 50 μM. Six of the compounds (including AG-690/13507628) showed detectable activities. The IC₅₀ values of four out of the six compounds were determined as AG-690/13507628 (2.806 μM), AG-690/13507724 (2.961 μM), AG-690/13507750 (19.19 μM), AG-690/13508164 (16.905 μM) (Figure 2; Table 2). The decreased IC50 of AG-690/13507628 from 9.457 μM to 2.806 μM in the second measurement is likely due to different experimental setup.

Figure 5.

The stacked bars graph for free energy contributions of the dominant residues of M^pro in binding to four compounds.

Table 2.

The structures and IC₅₀ of the four active compounds.

Compound	Specs ID	2D-structure	IC50 (μM)
1	AG-690/13507628		2.806 ± 0.642
2	AG-690/13507724		2.961 ± 0.636
3	AG-690/13507757		19.19 ± 1.94
4	AG-690/13508164		16.905 ± 2.795

Cytotoxicity

We next determined the cytotoxicity of the four active compounds (Figure 3). Due to different solubility of the compounds, the concentration of dimethyl sulfoxide (DMSO) and dimethylformamide (DMF) was 10‰ for AG-690/13507724 and 4‰ for the other three compounds. Compound AG-690/13507628, AG-690/13508164, and AG-690/13507724 have little effect on cell viability, but compound AG-690/13507757 has a greater inhibitory effect on cell growth. Considering that the IC₅₀ of AG-690/13507628 and AG-690/13507724 are 2.806 μM and 2.961 μM, respectively, these two compounds may hold promise for further optimizations.

Computational alanine scanning analysis

We performed MD simulations for these four compounds and details of the simulations are described in the Method section. Each M^pro-compound complex was simulation for 100 ns with three independent replicas. The RMSDs of the compounds are between 2 to 4 Å in the simulations, suggesting they are overall stable (Figure S6). Our computational alanine scanning analysis shows that residues His41, Met49, Leu27 and Thr25 contribute dominantly to the binding as shown in Figures 4 and 5. These four residues are conservative in binding to all these four compounds, indicating similar binding mechanism.

Figure 6.

Interaction diagram of pocket residues (within 4 Å) with two best compounds AG-690/13507628 (a, c) and AG-690/13507724 (b, d). The figure was prepared with PyMol (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC) and LigPlot (Laskowski & Swindells, 2011).

Discussion

To gain insights of the interactions between the active compounds and M^pro, we used Autodock vina (Trott & Olson, 2010) to generate the complex structures of two compounds AG-690/13507628 and AG-690/13507724, whose complex structures was extracted from the last frame in the 100 ns trajectory, that have better activity than the other two (Figure 6). These two compounds form specific interactions with four key residues including His41, Met49, Leu27 and Thr25. The hydrophilic cyano group is exposed to the solvent, together with the amide group, form a lever-like structure so that the benzene ring connected to the 6th position of the pyridine ring is inserted stably into the deep cavity. The benzene ring fits into the groove between Met165 and His41, and form π-π stacking interaction with His41.

Decomposition of the binding free energy of AG-690/13507628 using ASIE reveals six hot-spot residues: His41, Met49, Met165, Leu27, Thr25 and Gln189 (Table 3), which is in agreement with the structural analysis above. Among these residues, His41 is the dominant contributing residue with binding energy contribution great than 1.5 kcal/mol, and the main interaction being the van der Waals energy.

Table 3.

Residue-specific binding free energy of compound AG-690/13507628 from ASIE.

Residue	ΔΔEvdw	ΔΔEele	ΔΔGB	ΔΔNP	ΔΔH	–TΔΔS	ΔΔG
189GLN	–3.81	–1.35	1.60	–0.28	–3.84	1.03	–2.81
41HIE	–2.50	–0.97	1.08	–0.10	–2.49	0.64	–1.84
49MET	–2.40	–0.06	0.68	–0.17	–1.94	0.80	–1.14
27LEU	–0.82	–0.03	–0.03	–0.06	–0.94	0.08	–0.87
165MET	–1.26	–0.42	0.57	–0.07	–1.18	0.62	–0.56
25THR	–0.98	–0.13	0.36	–0.14	–0.89	0.37	–0.51
Total	–18.30	–4.31	6.62	–1.09	–17.09	5.52	–11.57

Only residues with ΔΔG < –0.50 kcal/mol are listed and all value are in kcal/mol.

Furthermore, the activity data obtained in this study also allows side-by-side comparison of active and non-active compounds to reveal key determinants of binding strength. To this end, we first compare in detail two compounds AG-690/13507757 (IC₅₀ = 19.19 μM) and AG-690/13507750 (non-active) that have high structural similarity but distinct activities (Figure 7a; Table S1). On the benzene ring attached to the amide, the chlorine group is closer to the hydroxyl group of Thr25 (2.5 Å) when it is in the para-position (AG-690/13507757), which facilitates the formation of hydrogen bonds with the hydroxyl group. In contrast, when the chlorine group is in the inter-position (AG-690/13507750), it is away from Thr25 (3.0 Å), therefore decrease the inhibitory activity.

Figure 7.

Comparison of active and non-active compounds. (a) AG-690/13507757 (active, yellow sticks) and AG-690/13507750 (non-active, green sticks). (b) AG-690/13507724 (active, pink sticks), AG-690/13507754 (non-active, blue sticks) and AG-690/13507732(nonactive, red sticks).

In AG-690/13507724 (IC₅₀ = 2.961 μM), when the substituent adjacent to the benzene ring attached to the amide is methoxy, it is easy to have a stable polar interaction with the carboxamide of Asn142 (2.6 Å). However, when the substituent is a benzene ring (AG-690/13507754, non-active, Figure 7b; Table S2), it is prone to collide with the surrounding amino acids due to the large size of the benzene ring. In the non-active compound AG-690/13507732, the substituent is a trifluoromethyl group (Figure 7b; Table S1), which cannot interact strongly with Asn142 in a stable manner due to its increased distance with Asn142 (3.1 Å).

Conclusion

Although FDA has issued an emergency use authorization on vaccines against COVID-19, drugs for anti-SARS-Cov2 are still extremely important (Lau et al., 2020). In this study, we discovered four inhibitors of SARS-Cov2 M^pro. To assess the structural novelty of these inhibitors, we used Rdkit to calculate the Morgan fingerprint similarity of these four inhibitors with other known inhibitors, and the highest score was only 0.513 (Table S2). Three of these compounds have low toxicity and several modifiable sites for further optimization. For example, the three benzene rings are all easily modifiable sites and have a considerable modification space: the phenyl ring at the 6th position of the pyridine ring may be replaced with hydrophobic group to form hydrophobic interactions with residue Met165. We have additionally predicted the ADMET properties of the four compounds using ADMETlab 2.0 (Xiong et al., 2021) and included the data in Table S3. Our compounds show good absorption, distribution and Excretion properties, but less desirable LogS, LogD and LogP values, indicating directions of further optimization. These results may provide a basis for further development of anti-SARS-CoV-2 drugs.

Funding Statement

This work was supported by the National Natural Science Foundation of China (Grant 91753103, 31700646, 21933010), and the Natural Science Foundation of Shanghai (Grant 19ZR1473600).