Virtual screening identification of novel severe acute respiratory syndrome 3C-like protease inhibitors and in vitro confirmation

Graphical abstract

Novel potent inhibitors of 3C-like protease (3CL^pro) of severe acute respiratory syndrome associated coronavirus (SARS-CoV) were identified through structure-based virtual screening and validated the inhibitory activity in vitro.

Abstract

The 3C-like protease (3CL^pro) of severe acute respiratory syndrome associated coronavirus (SARS-CoV) is vital for SARS-CoV replication and is a promising drug target. Structure based virtual screening of 308 307 chemical compounds was performed using the computation tool Autodock 3.0.5 on a WISDOM Production Environment. The top 1468 ranked compounds with free binding energy ranging from −14.0 to −17.09 kcal mol⁻¹ were selected to check the hydrogen bond interaction with amino acid residues in the active site of 3CL^pro. Fifty-three compounds from 35 main groups were tested in an in vitro assay for inhibition of 3CL^pro expressed by Escherichia coli. Seven of the 53 compounds were selected; their IC₅₀ ranged from 38.57 ± 2.41 to 101.38 ± 3.27 μM. Two strong 3CL^pro inhibitors were further identified as competitive inhibitors of 3CL^pro with K_i values of 9.11 ± 1.6 and 9.93 ± 0.44 μM. Hydrophobic and hydrogen bond interactions of compound with amino acid residues in the active site of 3CL^pro were also identified.

Severe acute respiratory syndrome (SARS) is a severe febrile respiratory illness caused by SARS-associated coronavirus (SARS-CoV).1, 2 A global outbreak of SARS between March 2003 and July 2003 caused over 8 000 probable or confirmed cases and 774 deaths.³ A mortality rate as high as 10% has been estimated by the World Health Organization.3, 4 The possibility of the re-emergence of SARS is a serious threat, since efficient therapy and a vaccine are not currently available.⁵ The SARS-CoV genome contains 14 functional open reading frames (ORFs).⁶ Two large 5′-terminal ORFs, designated 1a and 1b, encode two overlapping polyproteins, respectively designated pp1a and pp1b, which have to be cleaved extensively to produce proteins necessary for viral RNA synthesis and genome replication.6, 7 SARS-CoV 3C-like protease (3CL^pro) plays a major role in the processing of the viral polyproteins and control of the activity of the replicase complex.⁸ The enzymatic activity of 3CL^pro is essential for the viral life cycle and, therefore, represents an attractive target for the development of antiviral drugs directed against SARS-CoV and other coronavirus infections.⁹

Over the past decade, high-throughput virtual screening (VS) has emerged; especially structure based virtual screening (SBVS), as a reliable, cost-effective and time-saving technique for the discovery of lead compounds as an alternative to high-throughput screening.¹⁰ VS applied to the discovery of new enzyme inhibitors involves docking, computational fitting of structures of compounds to the active site of an enzyme, and scoring and ranking of each compound.¹¹

In this study, we identified novel 3CL^pro inhibitors among 308,307 compounds by SBVS with docking calculations on a WISDOM (Wide In Silico Docking On Malaria) production environment (WPE).¹² Additionally, the inhibition activity and enzyme kinetics were characterized by in vitro enzyme assay. A 3-Dimensional coordinate in the X-ray crystal structure of 3CL^pro (PDB accession code 2ZU5)¹³ obtained from the Protein Data Bank (PDB; http://www.pdb.org)¹⁴ was used as the receptor model in the SBVS with docking simulations. After removing the water molecules, the polar hydrogen (H) atoms were added to the macromolecule, histidine residues were made neutral and Kollman charges were assigned for all atoms.¹⁵ The docking library for 3CL^pro was comprised of 308 307 compounds from the Chembridge database (http://www.chembridge.com). All compounds were filtered to remove unsuitable components that would not reach and pass clinical trials due to undesired and toxic properties. The three dimensional atomic coordinates of the compounds included in the docking library were generated by the Corina program (Molecular Networks GmbH, Erlangen, Germany). AutoDock version 3.0.5 was used for the computational molecular docking simulation of flexible small molecules to rigid proteins with ligand and rigid proteins.¹⁶ Large scale computations were conducted between 2ZU5 and 308 307 compounds using the KISTI grid infrastructure.¹⁷ In the first Postdocking filtering strategy based on the free binding energy of the lowest energy conformation, 1468 top ranked compounds having a free binding energy ranging from −14.0 to −17.09 kcal mol⁻¹ were selected. To further narrow drug candidate selection, the Chimera software 1.4.1 program (University of California, San Francisco)¹⁸ was used to identify potential H-bonds between residues in the active site pocket of 3CL^pro. Selected compounds were analyzed for their hydrophobic and H-bond interactions using the Ligplot program.¹⁹ Among 1468 compounds, 22 compounds (1.5% of the top scoring compounds) exhibited no potential hydrogen bond (H-bond) interactions and 381 compounds (25.95% of the top scoring compounds) showed weak H-bond interactions with amino acid residues in the active site pocket of 3CL^pro. According to their free binding energy and H-bond interactions with key residues, 214 compounds were selected and classified into 35 groups by library MCS 0.7 (ChemAxon, San Francisco, CA, USA). Fifty-three compounds were selected from the 35 main clusters for in vitro inhibitory activity against 3CL^pro. The compounds provided by the vendor were each >90% pure and were used without further treatment.

The gene encoding 3CL^pro from SARS-CoV polyprotein (amino acid residues 3241–3546, GenBank accession no. AY274119) was constructed by a custom gene synthesis service (GenScript, Piscataway, NJ, USA). The 3CL^pro enzyme was expressed in E. coli BL21 (DE3) and purified using a Ni-Sepharose resin (GE Healthcare, Buckinghamshire, UK). The K _m value of 3CL^pro calculated from the double reciprocal Lineweaver–Burk plot by fluorogenic peptide Dabcyl-KTSAVLQSGFRKME-Edans using a SpectraMax Gemini XPS apparatus (Molecular Devices, Eugene, OR, USA) with excitation and fluorescence emission wavelengths of 355 and 538 nm, respectively. Each compound from VS was tested in duplicate at a concentration of 100 μM for its ability to inhibit 3CL^pro activity. Seven compounds showed optimal inhibition 3CL^pro at 100 μM (Table 1 ). The chemical structure of each compound is depicted in Figure 1 , with physicochemical and chemical names of each compound is supplied in the Supplementary data. The seven compounds displayed 3CL^pro inhibitory activities with IC₅₀ values ranging between 38.57 ± 2.41 and 101.38 ± 3.27 μM (Table 1). These compounds can be considered as a new inhibitor scaffold for further development by structure activity relationship or de novo design studies. Compounds 6 and 7, which were similar in structure and which displayed more than 60% inhibition activity at 100 μM, were subjected to kinetic characterization. To elucidate the inhibitory mechanism of both compounds, inhibition kinetic experiments were performed at different constant inhibitor concentrations (0–60 μM) and different substrate concentrations. For analysis of the modes of inhibition of these compounds, both Lineweaver–Burk and Dixon plots were used. A K _i value was calculated from the Dixon plots using the SigmaPlot program (SPSS, San Diego, CA, USA). Both compounds were competitive inhibitors. The inhibition of 3CL^pro by compound 7 is illustrated in Figure 2 . The Lineweaver–Burk plot of 1/v versus 1/[S] resulted in a family of straight lines with the same y-axis intercept reflecting competitive inhibition toward 3CL^pro (Fig. 2A). This activity suggests that compound 7 potently impairs the catalytic activity of 3CL^pro by binding in the enzyme’s catalytic site. The K _i value of compounds 6 and 7 were determined to be 9.11 ± 1.61 and 9.93 ± 0.44 μM, respectively, from the common x-axis intercept of lines on the corresponding Dixon plot (Fig. 2B). Compound 7 was analyzed by molecular docking as a potent binder to the active site pocket of 3CL^pro (Fig. 3 A). The predicted free binding energy between compound 7 and 3CL^pro is shown in Table 1. Figure 3B provides the details of the specific interactions between compound 7 and 3CL^pro; carbon atoms of compound 7 formed hydrophobic interactions with His41, Leu141, Phe140, Cys145, Glu166, His163, Gly170, and His172 of 3CL^pro. The O² atom of methylbenzamide group of compound 7 accepted H-bonding with the side chain carboxamide of Asn142 with a distance of 3.15 Å (Fig. 3B). The N² atom of the methacrylamide group of compound 7 donated 2.66 and 2.73 Å H-bond to the carboxyl group of Phe140 and the side chain carbonyl group of Glu166. The O⁴ atom of the nitrophenyl group of compound 7 formed 2.65 Å H-bonds with the backbone N atom of Gly143. The O¹ atom of the nitrophenyl group of compound 7 has two H-bonds: one H-bond with side chain of Cys145 with distance 3.23 Å and another one with the N atom of the main chain of Cys145 with a distance 3.23 Å. Viewing the H-bond between compound 7 and the amino acid residues in the active site of 3CL^pro revealed that compound 7 bound to the S1 site (the substrate binding pocket) of SARS-CoV through H-bonds with Phe140, Gly143, Cy145, and Glu166. S1 of SARS-CoV 3CL^pro confers absolute specificity for the Gln-P1 substrate residue.²⁰ The nitrophenyl group of compound 7 is very likely crucial in the 3CL^pro inhibition activity, given its H-bond formation with Cys145 and Gly143, as well as hydrophobic interactions with His41, Cys145 and Cys145. Hence, one of the catalytic dyad residues of 3CL^pro is vital for the binding of compound 7 to 3CL^pro.

Table 1.

Inhibitory activities of the identified compounds against recombinant 3CL^pro

Compounds	Chembridge ID	Free binding energy (kcal mol−1)	Inhibitiona (%)	IC50 (μM)
1	7007610	−14.5	58.23	58.35 ± 1.41
2	6939388	−15.09	61.36	62.79 ± 3.19
3	7742065	−15.17	49.14	101.38 ± 3.27
4	5623347	−15.20	56.11	77.09 ± 1.94
5	7935169	−15.75	53.27	90.72 ± 5.54
6	5874274	−15.02	82.59	38.57 ± 2.41
7	7112399	−15.13	81.43	41.39 ± 1.17

^aInhibition by 100 μM.

Figure 1.

Chemical structure of the novel 3CL^pro inhibitors.

Figure 2.

Lineweaver–Burk plot (A) and Dixon plot (B) analyses for the inhibition of 3CL^pro by compound 7. The kinetic constants, K_m and K_i, were calculated using linear regression analysis. (A) Compound 7 concentration of 0 μM (●), 20 μM (○), 30 μM (▾), 40 μM (Δ), 50 μM (■), and 60 μM (□). (B) FRET substrate concentration of 5 μM (●), 7.5 μM (○), and 10 μM (▾).

Figure 3.

Computational docking, hydrophobic and H-bond interactions of compound 7 with residues in the active site pocket of 3CL^pro. (A) Comparison of binding modes of compound 7 (magenta) in the active site pocket of 3CL^pro. (B) Hydrophobic and H-bond interactions of compound 7 with amino acid residues in the active site of 3CL^pro. H-Bond interactions are represented by green dashed lines.

In summary, VS of 308,307 compounds was done to identify novel 3CL^pro inhibitors. Fifty-three compounds were tested for their inhibitory activity towards 3CL^pro expressed from E. coli. The IC₅₀ of seven especially potent compounds ranged from 38.57 ± 2.41 to 101.38 ± 3.27 μM, and two of these compounds (6 and 7) were competitive inhibitors of 3CL^pro with K _i values of 9.11 ± 1.61 and 9.93 ± 0.44 μM. Detailed docking simulation binding mode analyses showed that the inhibitors could be stabilized by the formation of H-bonds with catalytic residues and the establishment of hydrophobic contacts at the opposite regions of the active site. More detailed inhibition kinetics and molecular modeling studies are underway to elucidate the inhibitory mechanism of compounds 6 and 7.