Discovery of potential FDA-approved SARS-CoV-2 Papain-like protease inhibitors: A multi-phase in silico approach

Abstract

Papain-like protease (PLpro) is a crucial enzyme for SARS-CoV-2 replication and immune evasion. Inhibiting PLpro could be a promising strategy to fight against COVID-19. This study aimed to identify potent inhibitors of PLpro among FDA-approved drugs using an in silico approach. The study also aimed to examine and confirm the binding of the selected compounds to the active pocket of PLpro using a multi-phased in silico approach, involving the screening of 3009 FDA-approved drugs to pinpoint the most similar compounds to, TTT, the co-crystallized ligand TTT of PLpro. The selected compounds were subjected to further analysis, including molecular docking, molecular dynamics simulations, MM-GPSA (molecular mechanics generalized Born surface area), and PLIP (Protein-Ligand Interaction Profiler) studies, to examine and confirm their binding to the active pocket of PLpro. Seven candidates (Vismodegib, Celecoxib, Ketoprofen, Indomethacin, Naphazoline, Valdecoxib, and Eslicarbazepine) showed promising in silico activities against the PLpro. The computational analysis confirmed the binding of Celecoxib to the active pocket of PLpro, suggesting its potential in the fight against COVID-19. This study identified seven FDA-approved drugs as potential inhibitors of PLpro, providing a feasible approach for drug repurposing against COVID-19. The results obtained from the in silico approach hold promise, but further in vitro and in vivo studies are warranted to validate the potential of these compounds.

Keywords

docking FDA-approved drugs MD simulations MM-GPSA PLIP SARS-CoV-2 papain-like protease structural similarity

A multi-phased in silico approach selected the most favorable FDA-approved drugs against SARS-CoV-2 PLpro. The study also examined and confirmed the binding of the selected compounds to the PLpro’s active pocket.

Introduction

The drugs that have been granted FDA approval have undergone rigorous evaluation by the Center for Drug Evaluation and Research (CDER) to assess their therapeutic benefits as well as potential adverse effects.¹ The FDA’s approval of a drug serves as an indication of its efficacy and overall safety profile.² Therefore, if a new use or indication for these drugs is discovered, they can be safely repurposed.³ The discovery of new bioactive molecules is a costly and lengthy process, typically taking around 12 years and costing about US$2.6 billion per drug,⁴ However, drug repositioning offers an efficient alternative for identifying and developing drugs with new pharmacological indications.⁵ Drug repositioning controls and decreases the timeline of drug development, as the used drug exhibited good ADMT properties in humans.⁶ Consequently, no need to conduct Phase I clinical trials.⁷ The repurposing process was frequently used in the revelation of several drugs with promising activities against cancer,^8,9 SARS-CoV-2,¹⁰ inflammations¹¹ bacterial,¹² parasitic,¹³ and viral¹⁴ infections.

Computer-based (in silico) chemistry approaches have been extensively utilized in pharmacology development and testing over the past 20 years.^15,16 This vast utilization was a result for various reasons. First, the identification of the exact 3D structures of various protein targets in the body of human.^17,18 Second, the recent tremendous progress in the scope of computers software as well as hardware.¹⁹ Third, the advancements in the understanding of the Structure Activity Relationship (SAR) and QSAR principles.²⁰ Computational chemistry encompasses two main approaches: ligand-based and structure-based. Ligand-based approaches, rely on the chemical structure of molecules that bind to the target of interest to predict the activity of new compounds. These methods allow to identify and optimize potential drug candidates based on their structural similarity to known active compounds such as structural similarity,²¹ molecular fingerprinting,²² Q-SAR,²³ pharmacophore,²⁴ ADMET properties,²⁵ and homology modeling.²⁶ However, structure-based approaches utilize the 3D structure of the target protein to identify and optimize binding interactions with potential ligands. These methods enable the visualization of how compounds interact with specific protein targets and aiding in the development of highly specific and potent drugs such as molecular docking,²⁷ and MD simulations.²⁸

To tackle the COVID-19 pandemic, our team employed various in silico techniques to identify natural inhibitors. From a collection of 310 natural antiviral metabolites, we pinpointed potential inhibitors for key SARS-CoV-2 proteins, including the main protease in silico(M^pro),^29,30 nsp10,³¹ helicase,³² and papain-like protease (PLpro).³³ In addition, from 3009 FDA-approved drugs, the most potent inhibitors for the SARS-CoV-2 RdRp,³⁴ M^pro,³⁵ and the SARS-CoV-2 nsp16-nsp10 2′-O-Methyltransferase Complex³⁶ were selected. Also, the potential inhibitors of SARS-CoV-2 RdRp,³⁷ helicase,³⁸ and PLpro³⁹ were selected from 4924 natural African metabolites. Between 5956 of traditional Chinese medicine metabolites, the potential natural SARS-CoV-2 Helicase inhibitors were determined.⁴⁰

The PLpro is a key enzyme that helps the virus replicate and evade the immune system. It cuts the viral polyprotein to produce necessary non-structural proteins and removes ubiquitin from host proteins to weaken the immune response.⁴¹ Because of its crucial role in the virus’s life cycle and immune suppression, PLpro is an important target for developing antiviral drugs. Inhibiting PLpro can stop the virus from replicating and boost the immune response, making these inhibitors potential treatments for COVID-19.⁴²

We herein report the preference of the most effective of SARS-CoV-2 PLpro (PDB ID: 3E9S) within 3009 clinical FDA-approved drugs.⁴³ The explored compounds’ chemical structures were retrieved from https://www.selleckchem.com/screening/fda-approved-drug-library.html.

TTT, a ligand with a strong binding affinity to the protein under study, was chosen as a reference compound. Various ligand-based (molecular fingerprints and similarity) and structure-based (molecular docking, MD simulations, MMGBSA and PLIP analysis) approaches have been utilized.

Results and discussion

Filter using molecular fingerprints

A co-crystallized ligand refers to a compound (ligand) that demonstrates a remarkable binding affinity with the associated protein, leading to the formation of a crystallizable ligand-protein complex.⁴⁴ This complex provides valuable insights into the interactions between the ligand and protein, shedding light on the key structural and chemical features that contribute to the optimum binding affinity. Consequently, the co-crystallized ligand’s chemical structure serves as a valuable model for designing and developing inhibitors that can effectively bind to the target protein.⁴⁵ By studying the structural characteristics and functional groups present in the co-crystallized ligand, we can gain a deeper understanding of the key factors responsible for its strong binding affinity. We used this knowledge to guide the selection of compounds with structural similarities to TTT, aiming to discover potent inhibitors with a high affinity for the PLpro protein. This approach is grounded in the basic principle of TTT Structure-Activity Relationship (SAR), which states that similarities in chemical structure between two compounds are likely to result in similarities in biological activity.⁴⁶ The ligand-based in silico methods can anticipate the biological effect of a compound depending on the chemical structure method based scientifically on the SAR principles.⁴⁵

Molecular fingerprints study is a kind of ligand-based computer-based in silico methods. The utilization of molecular fingerprints involves establishing the degree of similarity between the chemical structures of the compounds under investigation.⁴⁷ The molecular fingerprints experiment outlined the presence and or absence of the below features in atoms and fragments of the investigated compounds and TTT: H-bond accepting,⁴⁸ H-bond donating,⁴⁹ charged atoms,⁵⁰ hybridization,⁵¹ positively ionized atoms,⁵² negatively ionized atoms,⁵³ halogens,⁵⁴ aromatics,⁵⁵ as well as the ALogP.⁵⁶ Table 1 shows the most similar 150 candidates to TTT as well as the shared fingerprints.

Table 1.

Shared and absent structural fingerprints between the tested FDA-approved compounds and TTT.

Comp.	Similarity	S-A	S-B	S-C	Comp.	Similarity	S-A	S-B	S-C
TTT	1	454	0	0	1155	0.739216	377	56	77
1	0.775	431	102	23	1156	0.710956	305	-25	149
12	0.769	370	27	84	1159	0.703704	323	5	131
14	0.710909	391	96	63	1174	0.7343	304	-40	150
19	0.786	431	94	23	1181	0.707661	351	42	103
23	0.769	399	65	55	1218	0.725877	331	2	123
24	0.746637	333	-8	121	1286	0.72104	305	-31	149
65	0.758	323	-28	131	1287	0.782	374	24	80
69	0.803	453	110	1	1301	0.782	456	129	-2
70	0.727455	363	45	91	1307	0.711316	308	-21	146
85	0.704082	345	36	109	1377	0.704545	434	162	20
86	0.809	444	95	10	1427	0.743119	324	-18	130
109	0.728477	330	-1	124	1453	0.718535	314	-17	140
144	0.701923	438	170	16	1507	0.703196	308	-16	146
174	0.733432	498	225	-44	1511	0.715859	325	0	129
175	0.737271	362	37	92	1547	0.757	492	196	-38
183	0.734483	426	126	28	1565	0.728938	398	92	56
205	0.717728	417	127	37	1659	0.705263	335	21	119
209	0.710407	471	209	-17	1705	0.719457	318	-12	136
215	0.722376	523	270	-69	1706	0.70852	316	-8	138
216	0.744722	388	67	66	1709	0.784	484	163	-30
240	0.752834	332	-13	122	1711	0.752538	593	334	-139
246	0.719715	303	-33	151	1726	0.726547	364	47	90
249	0.731392	452	164	2	1738	0.708861	336	20	118
277	0.71319	465	198	-11	1751	0.713472	376	73	78
284	0.715517	332	10	122	1809	0.765	440	121	14
287	0.702461	314	-7	140	1834	0.742794	335	-3	119
353	0.734831	327	-9	127	1850	0.742515	496	214	-42
367	0.726862	322	-11	132	1870	0.785	362	7	92
387	0.704028	402	117	52	1879	0.765	417	91	37
418	0.734558	440	145	14	1880	0.715415	362	52	92
426	0.729258	334	4	120	1915	0.744035	343	7	111
451	0.717778	323	-4	131	1934	0.731602	338	8	116
500	0.761	351	7	103	1956	0.723881	388	82	66
548	0.717822	290	-50	164	2010	0.73774	346	15	108
550	0.715481	342	24	112	2033	0.708696	326	6	128
551	0.74569	346	10	108	2034	0.723164	512	254	-58
564	0.702317	394	107	60	2035	0.725838	368	53	86
592	0.721781	454	175	0	2073	0.808	365	-2	89
619	0.727273	512	250	-58	2078	0.725843	323	-9	131
632	0.708333	306	-22	148	2102	0.708763	275	-66	179
639	0.716707	296	-41	158	2170	0.7119	341	25	113
640	0.777	466	146	-12	2175	0.724771	316	-18	138
674	0.82	451	96	3	2178	0.762	342	-5	112
698	0.712215	344	29	110	2180	0.738977	419	113	35
699	0.74103	475	187	-21	2204	0.742597	652	424	-198
731	0.78	365	14	89	2265	0.725352	412	114	42
746	0.711364	313	-14	141	2297	0.730689	350	25	104
749	0.724744	495	229	-41	2308	0.766	397	64	57
750	0.733781	328	-7	126	2310	0.720532	379	72	75
766	0.718248	492	231	-38	2321	0.757	467	163	-13
767	0.711538	333	14	121	2330	0.730845	372	55	82
783	0.744348	428	121	26	2336	0.737274	449	155	5
806	0.73221	391	80	63	2345	0.729452	426	130	28
813	0.719643	403	106	51	2370	0.7119	341	25	113
817	0.838	420	47	34	2449	0.708779	331	13	123
827	0.763	393	61	61	2504	0.701754	480	230	-26
859	0.742798	361	32	93	2555	0.721338	453	174	1
860	0.727428	427	133	27	2625	0.763	355	11	99
866	0.741546	307	-40	147	2650	0.845	451	80	3
895	0.718341	329	4	125	2651	0.723849	346	24	108
896	0.722793	352	33	102	2667	0.741768	428	123	26
904	0.783	371	20	83	2701	0.735808	337	4	117
908	0.706597	407	122	47	2730	0.723502	471	197	-17
921	0.742317	314	-31	140	2741	0.740175	339	4	115
922	0.712297	307	-23	147	2743	0.737819	318	-23	136
964	0.713004	318	-8	136	2762	0.708211	483	228	-29
1002	0.714964	301	-33	153	2833	0.752834	332	-13	122
1024	0.716279	308	-24	146	2836	0.865	453	70	1
1036	0.76	494	196	-40	2841	0.720554	312	-21	142
1051	0.710359	336	19	118	2842	0.718341	329	4	125
1074	0.750416	451	147	3	2847	0.740558	451	155	3
1090	0.708241	318	-5	136	2852	0.703448	306	-19	148
1135	0.776	332	-26	122	2887	0.734694	324	-13	130
1148	0.711009	310	-18	144	2903	0.733333	330	-4	124
1149	0.733781	328	-7	126

S-A: The number bits are common in TTT and the FDA-approved drug.

S-B: The number bits in the FDA-approved drug but not TTT.

S-C: The number bits in TTT but not the FDA-approved drug.

Molecular Similarity

One key difference between molecular similarity and fingerprint studies is the level of molecular detail they capture. Molecular similarity studies typically consider a broader array of molecular descriptors and properties, allowing for a more thorough evaluation of structural and chemical similarities. These studies can account for factors such as shape, electrostatics, and pharmacophoric features. In contrast, fingerprint studies focus on capturing and comparing specific structural patterns encoded in binary fingerprints, offering a more concise representation of molecular structures.⁵⁷ Molecular similarity studies describe and compare the entire structures of the reference compound and the test set, using descriptors that are steric, electronic, topological, or physical.⁵⁸ To select the most similar 30 compounds to TTT (Figure 1 and Table 2), a molecular structure similarity has been conducted to compute the minimum distance between the examined compounds and TTT considering the following physiochemical features and structural descriptors; hydrogen bond donors (HBA),⁵⁹ acceptors (HBD)⁶⁰ partition coefficient (ALog p),⁶¹ molecular weight (M. W),⁶² rotatable bonds (RB),⁶³ cyclic rings (R) as well as aromatic rings (AR)⁶⁴ and the molecular fractional polar surface area (MFPSA).⁶⁵

Figure 1.

Molecular similarity analysis of the FDA-approved drugs and TTT.

Table 2.

Molecular structural properties of the FDA-approved drugs and TTT.

Comp.	ALog p	M. W	HBA	HBD	RB	R	AR	MFPSA	Minimum Distance
TTT	3.65	304.39	2	2	3	3	3	0.171	0
23	4.27	421.3	4	1	4	3	3	0.219	0.776
70	4.43	381.37	3	1	4	3	3	0.244	0.623
240	3.36	254.28	3	1	4	2	2	0.207	0.785
284	4.24	357.79	4	1	4	3	3	0.195	0.619
548	2.44	246.74	2	1	2	3	2	0.095	0.769
551	2.11	252.27	2	2	2	3	2	0.243	0.735
632	4.65	321.16	3	1	3	3	3	0.189	0.475
639	4.16	261.7	3	2	3	2	2	0.193	0.693
731	4.23	333.88	2	1	6	3	3	0.142	0.625
859	2.6	314.36	3	1	3	3	3	0.313	0.694
866	3.98	241.29	3	2	3	2	2	0.192	0.712
904	3.59	249.26	3	1	2	3	3	0.205	0.489
1002	3.68	244.26	2	1	3	2	2	0.149	0.742
1024	3.95	281.23	3	2	4	2	2	0.188	0.685
1149	3.06	295.29	4	2	4	3	3	0.286	0.628
1155	3.26	370.42	4	1	5	3	3	0.27	0.77
1159	2.4	296.32	3	1	2	3	2	0.244	0.76
1218	2.62	320.39	3	2	3	4	3	0.228	0.565
1286	2.62	283.3	3	1	2	3	2	0.205	0.704
1511	3.16	295.33	3	1	4	3	2	0.193	0.649
1870	2.32	317.34	3	2	2	4	3	0.247	0.655
2073	3.85	297.42	2	1	6	3	3	0.161	0.588
2178	3.27	282.29	4	1	5	3	3	0.229	0.665
2308	2.55	319.4	2	1	6	3	3	0.176	0.658
2336	4.01	318.32	4	2	2	4	3	0.227	0.637
2651	4.76	355.86	4	2	6	3	3	0.131	0.756
2701	4	280.32	3	1	3	3	2	0.167	0.624
2833	3.36	254.28	3	1	4	2	2	0.207	0.785
2842	2.96	323.36	2	3	3	4	3	0.178	0.543
2887	3.31	208.26	1	1	2	3	3	0.134	0.601

The results demonstrated a high degree of similarity between TTT and 30 FDA-approved compounds. The chemical structures of the selected compounds were illustrated in Figure 2.

Figure 2.

The chemical structures of the most like 30 FDA-approved compounds and TTT.

Docking studies

The most similar drugs with the co-crystallized ligand (TTT) of PLpro (PDB ID: 3E9S) were subjected to the docking studies against such crystal structure of protein. These compounds include 23, 70, 240, 284, 548, 551, 632, 639, 731, 859, 866, 904, 1002, 1024, 1149, 1155, 1218, 1286, 1511, 1870, 2073, 2178, 2308, 2336, 2651, 2701, 2833, 2842, and 2887.

In these studies, it was depended on the binding mode of the co-crystallized ligand as a guide and reference. Table 3 summaries the binding energy of the tested compounds against PLpro (PDB ID: 3E9S).

Table 3.

The calculated ∆G of the FDA-approved drugs and TTT.

Comp.	Name	∆G [Kcal/mole]
TTT		-21.81
23	Vismodegib	-13.78
70	Celecoxib	-17.85
240	ketoprofen	-15.88
284	indomethacin	-7.08
548	Naphazoline	-9.70
551	Phenytoin	-8.37
632	Lonidamine	-10.56
639	Tolfenamic acid	-13.16
731	Duloxetine	-13.60
859	Valdecoxib	-10.31
866	Mefenamic acid	-13.03
904	Cinchophen	-10.92
1002	(S)-(+)-FLURBIPROFEN	-12.14
1024	Flufenamic acid	-14.76
1149	Mebendazole	-16.98
1155	Parecoxib	-7.49
1159	Eslicarbazepine	-8.94
1218	Niraparib	-16.46
1286	Afloqualone	-7.78
1511	Indobufen	-8.47
1870	CP21R7 (CP21)	-11.54
2073	Duloxetine	-11.87
2178	Bendazac	-15.35
2308	Imidafenacin	-11.23
2336	Phenolphthalein	-9.63
2651	Amodiaquine	-14.05
2701	Phenprocoumon	-9.86
2833	S-(+)-Ketoprofen	-9.85
2842	Rucaparib	-15.32
2887	Bendazol	-13.40

To confirm the accuracy of our docking processes, we began with a validation step. This involved docking TTT into the active site of the PLpro (PDB ID: 3E9S) and then calculating the RMSD. The results showed an RMSD value of 1.2 Å between the original and docked poses. As previously reported, an RMSD below 2.0 Å is regarded as low and acceptable^66,67 demonstrating the high efficiency of our docking process (Figure 3).

Figure 3.

Superimposition of the docked form of TTT with its original form showing RMSD value of 1.2 Å.

The co-crystallized ligand (TTT) showed docking score value of -21.81 kcal/mol. It occupied the sub-pockets of the active site forming a hydrophobic interaction with Asp165 in addition to two hydrogen bonds with Asp165 and Gln270 (Figure 4).

Figure 4.

3D interaction of TTT in the active site of SARS papain-like protease.

The molecular docking studies revealed that seven drugs (Vismodegib 23, Celecoxib 70, Ketoprofen 240, Indomethacin 284, Naphazoline 548, Valdecoxib 859, and Eslicarbazepine 1159) exhibited good binding energies against the PLpro with binding modes like TTT. These drugs occupied the sub-pockets of the active site forming hydrogen bonds and hydrophobic interaction with the crucial amino acids in the PLpro active site. Table 4 summarizes the binding mode of the most promising drugs showing the numbers of the hydrogen and hydrophobic bonds as well as the names of amino acids incorporated in the binding. The binding modes of compounds 23 (Figure 5(a)), 70 (Figure 5(b)), 240 (Figure 5(c)), 284(Figure 5(d)), 548 (Figure 5(e)), 859 (Figure 5(f)), and 1159 (Figure 5(g)), were illustrated in 3D forms.

Table 4.

Summary of the binding modes of the tested compounds and TTT.

Comp.	Hydrogen bonds	Amino acid involved	Hydrophobic bonds	Amino acid involved
TTT	2	Asp165, Gln270	1	Asp165
23	3	Asp165, Lys158, Gln270	–	–
70	3	Asp165, Gln270, Leu163	1	Arg167
240	4	Asp165, Gln270, Tyr269	1	Arg167
284	3	Asp165, Lys158, Gln270	–	–
548	2	Asp165, Gln270	1	Tyr256
859	3	Asp165, Gln270, Arg167	1	Asp165
1159	2	Asp165, Gln270	2	Arg167
2833	4	Asp165, Gln270, Arg167	–	–

Figure 5.

3D interactions of 23 (a), 70 (b), 240 (c), 284 (d), 548 (e), 859 (f), and 1159 (g) in the active site of PLpro.

MD simulations studies

Molecular docking is an in silico technique that reveals the binding site of a ligand inside of a protein structure as a result of in silico experiments. A docking study has the disadvantage of describing proteins' interactions as rigid entities. The conformational changes of a protein's structure after binding with an active compound are therefore not considered during docking experiments.⁵⁵ However, molecular dynamics (MD) simulations serve as a robust computational tool for investigating and comprehending protein dynamics and structural modifications at the atomic resolution.⁶⁸ By accurately representing the physical interactions of individual atoms within the targeted protein, MD simulations enable the exploration of conformational changes that occur upon ligand binding. Consequently, it provides valuable insights into the dynamic and energetic behaviors of that protein, shedding light on the mechanisms of ligand recognition and binding-induced structural alterations.⁵⁶

Upon investigation of the binding modes and binding energy of the tested compounds against SARS papain-like protease (PDB ID: 3E9S), it was found that celecoxib, 70, is the most promising member. It showed the highest binding energy among the tested compound. Similar to the co-crystallized ligand, Celecoxib formed essential H. bonding interactions with the two crucial amino acids (Asp165 and Gln270). In addition, it formed an H-I with Arg167 as the co-crystallized ligand. Furthermore, celecoxib formed an extra H. bond with Leu163. From these results, it can be deduced that celecoxib has a good binding affinity against PLpro. Accordingly, it was selected for MD simulations investigations. In accordance with the analysis carried out on the production run of the MD, the obtained complex showed a stable behavior for both the PLpro and celecoxib. The RMSD plot (Figure 6(a)) showed a stable average for the PLpro (blue) and the obtained complex (green) expressing RMSD values around 2.4 Å. Likely, the RMSD of celecoxib (red) shows stable values with small fluctuations having an average of 1.5 Å. The Radius of Gyration (RoG, Figure 6(b)), and Solvent Accessible Surface Area (SASA, Figure 6(c)) show stable trend. They have an average of 23.9 Å and 17145 Å² for the RoG and SASA, respectively. Fortunately, the H. bonds (Figure 6(d)) show a stable trend also with 82 bonds as an average. The fluctuation of PLpro amino acids was depicted in the Root-Mean-Square Fluctuation (RMSF) plot (Figure 6(e)) showing very low level of fluctuation (less than 2 Å) except for the free N-terminal, Val189:Gln195 loop, Ile223:Thr232 loop, and the C-terminal reaching 9.6 Å, 3.7 Å, 4.3 Å, and 4.4 Å, respectively. Celecoxib remained almost in its place relative to the PLpro center of mass (Figure 6(f)) showing an average of 17.25 Å during the simulation time.

Figure 6.

Different calculated measurements from the MD trajectories. (a) RMSD: PLpro (blue), celecoxib (red), and the complex (green), (b) RoG, (c) SASA, (d) Number of H. bonds over the 100 ns, (e) RMSF, and (f) Distance between the centers of mass of celecoxib and PLpro.

MM-GBSA calculations

The analysis of binding free energy was carried out by MM-GBSA (Figure 7) and showed the various components of energy that contributed to the binding of celecoxib B to the PLpro. Celecoxib showed a total exact binding energy with the average value of -19.04 Kcal/Mol. In addition, it was found that the van der Waals energy was the largest, and only, favorable contribution with an average value of -34.49 Kcal/Mol. TTT exhibited a precise binding energy total with an average of -29.31 Kcal/Mol. Furthermore, it was discovered that the van der Waals energy had an average of -34.55 Kcal/Mol, which was the same as Celecoxib’s average and was a beneficial contribution.

Figure 7.

Different energetic components of MM-GBSA. Bars represent the standard deviations.

Moreover, a decomposition analysis (Figure 8) was preceded to identify which amino acid residues within 10 Å of PLpro that had the highest contribution to the interaction with both celecoxib and TTT. Starting with TTT, five amino acids (TYR269, GLN270, TYR265, PRO248, and PRO249) have energies lower than -1 Kcal/Mol (Figure 8(a)). Similarly, in case of celecoxib (Figure 8(b)), three amino acids were shared with TTT (Tyr269 (-2.14 Kcal/Mol), Pro248 (-1.64 Kcal/Mol), and Pro249 (-1.23 Kcal/Mol)) in addition to Thr302 (-1.83 Kcal/Mol). In addition, there are one amino acid (Asp303) that has a large non favorable interaction with an average of +2.34 Kcal/Mol. Interestingly, as Figure 8 shows, the decomposition analysis shows that celecoxib and TTT had the same interactions with several other amino acids indicating the high degree of similarity in binding.

Figure 8.

Binding free energy decomposition of (a) the PLpro-TTT complex, and (b) PLpro-celecoxib complex.

PLIP investigation

To have a deeper insight, protein-ligand interaction profiler (PLIP) studies were preceded. The resulted trajectory from the MD was clustered and separated into representative frames for every produced cluster. The number of the resulted clusters was automatically selected by the elbow method producing five clusters. The PLIP webserver was utilized for every cluster to explore the types and number of interactions between celecoxib and PLpro. Table 5 represents the obtained types and number of those interactions based on the PLIP webserver. The major interaction that was detected is the hydrophobic interaction with 32 interactions in all cluster representatives. This aligns with the significant difference in van der Waals and electrostatic energy values derived from the MM-GBSA analysis. The most common amino acid is Tyr269 (eight hydrophobic interaction) followed by Asp165 (four hydrophobic interaction), and Arg167 (four hydrophobic interaction). Besides the production of interaction numbers and types from PLIP, it also generated a .pse file to see celecoxib’s 3D conformations as well as celecoxib’s interaction with PLpro (Figure 9).

Table 5.

Number and types of interactions identified for celecoxib using the PLIP webserver. The amino acids in bold are the most frequently occurring ones in all cluster representatives.

Cluster	Hydrophobic interactions	Amino acids in PLpro
C1	5	Asp165—Arg167—Pro249—Tyr265—Tyr269
C2	10	Leu163—Asp165—Arg167—Pro248—Pro249—Tyr265 (2)—Tyr269 (2)—Tyr274
C3	6	Asp165—Arg167—Tyr265—Tyr269 (2)—Gln270
C4	3	Leu163—Tyr269—Thr302
C5	8	Leu163—Asp165—Arg167—Pro249—Tyr265—Tyr269 (2)—Gln270

Figure 9.

PLIP interactions for each cluster representative for celecoxib. Hydrophobic interaction (dashed gray), amino acids (blue sticks), and celecoxib (orange sticks).

The PLIP webserver was utilized for every cluster to explore the types and number of interactions between TTT and PLpro. Table 6 represents the obtained types and number of those interactions based on the PLIP webserver. The primary interaction identified was the hydrophobic interaction, which was present in all cluster representatives totaling 41 interactions. This is consistent with the significant discrepancy in the van der Waals and electrostatic energy values derived from the MM-GBSA. Y265 is the most prevalent amino acid with 11 hydrophobic interactions, next is Y269 with 9 hydrophobic interactions, and P249 with 5 hydrophobic interactions. In addition to generating interaction numbers and types from PLIP, a .pse file was also created to visualize TTT’s 3D conformations and its interaction with PLpro (Figure 10).

Table 6.

Number and types of interactions identified for TTT using the PLIP webserver. The amino acids in bold are the most frequently occurring ones in all cluster representatives.

Cluster Number	Number of Hydrophobic Interactions	Amino Acids in PLpro	Number of Hydrogen Bonds	Amino Acids in PLpro
C1	8	D165—P249—Y265 (2)—Y269 (2)—Q270—Y274	2	Q270 (2)
C2	7	L163—P248—P249—Y265 (2)—Y269—Q270	1	Q270
C3	8	L163—D165—P249—Y265 (2)—Y269 (3)	5	Y265—Y269 (2)—Q270 (2)
C4	7	L163—P249—Y265 (2)—Y269—Y274—T302	2	Q270 (2)
C5	11	L163—D165—P249—Y265 (3)—Y269 (2)—Q270—Y274—T302	4	Y265—Y269—Q270 (2)

Figure 10.

PLIP interactions for each cluster representative for TTT. Hydrophobic interaction (dashed gray), amino acids (blue sticks), and TTT (orange sticks).

Conclusion

In conclusion, our study utilized a comprehensive in silico approach to identify potent inhibitors of the SARS-CoV-2 papain-like protease (PLpro) among 3009 FDA-approved drugs. Through rigorous screening and computational experiments, we identified seven promising candidates (Vismodegib, Celecoxib, Ketoprofen, Indomethacin, Naphazoline, Valdecoxib, and Eslicarbazepine) that demonstrated high structural similarity to the co-crystallized ligand (TTT) of PLpro in addition to good binding modes. Further analysis, including MD simulations, MM-GPSA, and PLIP studies, confirmed the binding of Celecoxib to the active pocket of PLpro. The results obtained from our study provide significant potential in the ongoing fight against COVID-19. These identified inhibitors hold promise for further development as therapeutic agents against SARS-CoV-2. However, it is crucial to emphasize that further in vitro and in vivo studies are necessary to validate the efficacy, and clinical potential of these compounds.

Method

Fingerprint studies

Discovery Studio 4.0 software was employed to perform a molecular fingerprint analysis on 3009 FDA-approved drugs, ultimately selecting the 150 most similar candidates to TTT. Detailed information on this process can be found in the methods section of the Supplementary Data.

Molecular Similarity detection

Discovery Studio 4.0 software was employed to perform a molecular similarity analysis on the selected 150 FDA-approved drugs, ultimately selecting the 30 most similar candidates to TTT. Detailed information on this process can be found in the methods section of the Supplementary Data.

Docking studies

Discovery studio software was employed to run the docking studies against the PLpro (PDB ID: 3E9S) for the most like 30 FDA-approved drugs (23, 70, 240, 284, 548, 551, 632, 639, 731, 859, 866, 904, 1002, 1024, 1149, 1155, 1218, 1286, 1511, 1870, 2073, 2178, 2308, 2336, 2651, 2701, 2833, 2842, and 2887) using TTT as a reference molecule.⁶⁹ Detailed information on this process can be found in the methods section of the Supplementary Data.

MD Simulations

PLpro-Celecoxib complex’s stability and structural changes were examined for 100 ns utilizing unbiased MD simulation in GROMACS 2021.^70–72 Input files were prepared using the CHARMM-GUI server.^73,74 Binding energy of the PLpro-Celecoxib and PLpro-TTT complexes were computed using the MM-GBSA method with the gmx_MMPBSA program.^75–77 Detailed information on this process can be found in the methods section of the Supplementary Data.

Supplemental Material

sj-docx-1-chl-10.1177_17475198241298547 – Supplemental material for Discovery of potential FDA-approved SARS-CoV-2 Papain-like protease inhibitors: A multi-phase in silico approach

Supplemental material, sj-docx-1-chl-10.1177_17475198241298547 for Discovery of potential FDA-approved SARS-CoV-2 Papain-like protease inhibitors: A multi-phase in silico approach by Ahmed M Metwaly, Eslam B Elkaeed, Mohamed M Khalifa, Aisha A Alsfouk, Fatma G Amin, Ibrahim M Ibrahim and Ibrahim H Eissa in Journal of Chemical Research

Footnotes

Acknowledgements

The authors thank AlMaarefa University, Riyadh, Saudi Arabia, for supporting this research

Author contributions

A.M.M. and I.H.E.: Supervision, conceptualization, and writing of the original draft. E.B.E.: Fingerprints and similarity analysis, writing, and manuscript revision. M.M.K.: Molecular docking studies. A.A.A.: Software development and support, writing, and manuscript revision. F.G.A.: MM-GBSA and PLIP calculations. I.M.I.: Molecular dynamics (MD) simulations. All authors revised and approved the final version of the manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by Princess Nourah bint Abdulrahman University Researchers Supporting Project number (PNURSP2024R116), Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia.

ORCID iD

Ahmed M Metwaly

Supplemental material

Supplemental material for this article is available online.

References

Ciociola

Cohen

Kulkarni

, et al. How drugs are developed and approved by the FDA: current process and future directions. Am Coll Gastroenterol 2014; 109(5): 620–623.

Brown

Wobst

. A decade of FDA-approved drugs (2010–2019): trends and future directions. J Med Chem 2021; 64(5): 2312–2338.

Pushpakom

Iorio

Eyers

, et al. Drug repurposing: progress, challenges and recommendations. Nature Rev Drug Discover 2019; 18(1): 41–58.

Chan

Shan

Dahoun

, et al. Advancing drug discovery via artificial intelligence. Trend Pharmacol Sci 2019; 40(8): 592–604.

Parvathaneni

Kulkarni

Muth

, et al. Drug repurposing: a promising tool to accelerate the drug discovery process. Drug Discovery Today 2019; 24(10): 2076–2085.

Oprea

Mestres

. Drug repurposing: far beyond new targets for old drugs. AAPS J 2012; 14(4): 759–763.

Oprea

Bauman

Bologa

, et al. Drug repurposing from an academic perspective. Drug Discover Today: Therapeutic Strat 2011; 8(3-4): 61–69.

Sleire

Førde

Netland

, et al. Drug repurposing in cancer. Pharmacol Res 2017; 124: 74–91.

Nowak-Sliwinska

Scapozza

i Altaba

. Drug repurposing in oncology: compounds, pathways, phenotypes and computational approaches for colorectal cancer. Biochim Biophys Acta Rev Cancer 2019; 1871(2): 434–454.

10.

Singh

Parida

Lingaraju

, et al. Drug repurposing approach to fight COVID-19. Pharmacol Rep 2020; 72(6): 1479–1508.

11.

Hong

Bang

. Anti-inflammatory strategies for schizophrenia: a review of evidence for therapeutic applications and drug repurposing. Clin Psychopharmacol Neurosci 2020; 18(1): 10.

12.

Konreddy

Rani

Lee

, et al. Recent drug-repurposing-driven advances in the discovery of novel antibiotics. Curr Med Chem 2019; 26(28): 5363–5388.

13.

Shirley

Sharma

Warren

, et al. Drug repurposing of the alcohol abuse medication disulfiram as an anti-parasitic agent. Front Cell Infect Microbiol 2021; 11: 165.

14.

Trivedi

Mohan

Byrareddy

. Drug repurposing approaches to combating viral infections. J Clin Med 2020; 9(11): 3777.

15.

Hagler

. Chemoinformatics and drug discovery. Molecules 2002; 7(8): 566–600.

16.

Engel

. Basic overview of chemoinformatics. J Chem Inform Model 2006; 46(6): 2267–2277.

17.

Westbrook

Burley

. How structural biologists and the Protein Data Bank contributed to recent FDA new drug approvals. Structure 2019; 27(2): 211–217.

18.

Eissa

Yousef

Sami

, et al. Exploring the anticancer properties of a new nicotinamide analogue: investigations into in silico analysis, antiproliferative effects, selectivity, VEGFR-2 inhibition, apoptosis induction, and migration suppression. Pathol Res Pract 2023; 252: 154924.

19.

Grimme

Schreiner

. Computational chemistry: the fate of current methods and future challenges. Angewandte Chemie Int Edn 2018; 57(16): 4170–4176.

20.

Amin

Banerjee

Sing

, et al. First structure–activity relationship analysis of SARS-CoV-2 virus main protease (Mpro) inhibitors: an endeavor on COVID-19 drug discovery. Molecul Divers 2021; 25(3): 1827–1838.

21.

Ranjan

Devarapalli

Kundu

, et al. Isomorphism: molecular similarity to crystal structure similarity in multicomponent forms of analgesic drugs tolfenamic and mefenamic acid. IUCrJ 2020; 7(2): 173–183.

22.

Baidya

Ghosh

Amin

, et al. In silico modeling, identification of crucial molecular fingerprints, and prediction of new possible substrates of human organic cationic transporters 1 and 2. New J Chem 2020; 44(10): 4129–4143.

23.

Shi

. Support vector regression-based QSAR models for prediction of antioxidant activity of phenolic compounds. Scientific Reports 2021; 11(1): 1–9.

24.

Idris

Yekeen

Alakanse

, et al. Computer-aided screening for potential TMPRSS2 inhibitors: a combination of pharmacophore modeling, molecular docking and molecular dynamics simulation approaches. J Biomol Struct Dynam 2021; 39(15): 5638–5656.

25.

Cheng

Liu

, et al. In silico ADMET prediction: recent advances, current challenges and future trends. Curr Med Chem 2013; 13(11): 1273–1289.

26.

. A new computer model for evaluating the selective binding affinity of phenylalkylamines to T-Type Ca2+ channels. Pharmaceuticals 2021; 14(2): 141.

27.

Fan

Zhang

LJQ

. Progress in molecular docking. Molecules 2019; 7: 83–89.

28.

Hospital

Goñi

Orozco

, et al. Molecular dynamics simulations: advances and applications. Ann Report Comput Chem 2015: 37–47.

29.

Elkaeed

Youssef

Eissa

, et al. Multi-step in silico discovery of natural drugs against COVID-19 targeting main protease. Int J Mol Sci 2022; 23(13): 6912.

30.

Elkaeed

Eissa

Elkady

, et al. A multistage in silico study of natural potential inhibitors targeting SARS-CoV-2 main protease. Int J Mol Sci 2022; 23(15): 8407.

31.

Eissa

Khalifa

Elkaeed

, et al. In silico exploration of potential natural inhibitors against SARS-CoV-2 nsp10. Molecules 2021; 26(20): 6151.

32.

Elkaeed

Eissa

Saleh

, et al. Computer-aided drug discovery of natural antiviral metabolites as potential SARS-CoV-2 helicase inhibitors. J Chem Res 2024; 48(1): 17475198231221253.

33.

Elkaeed

Metwaly

Alesawy

, et al. Discovery of potential SARS-CoV-2 papain-like protease natural inhibitors employing a multi-phase in silico approach. Life 2022; 12(9): 1407.

34.

Elkaeed

Elkady

Belal

, et al. Multi-phase in silico discovery of potential SARS-CoV-2 RNA-dependent RNA polymerase inhibitors among 3009 clinical and FDA-approved related drugs. J Chem 2022; 10(3): 530.

35.

Eissa

Saleh

Al-Rashood

, et al. Multistaged in silico discovery of the best SARS-CoV-2 main protease inhibitors amongst 3009 clinical and FDA-approved compounds. J Chem 2024; 2024: 5084553.

36.

Eissa

Alesawy

Saleh

, et al. Ligand and structure-based in silico determination of the most promising SARS-CoV-2 nsp16-nsp10 2′-O-methyltransferase complex inhibitors among 3009 FDA-approved drugs. J Chem 2022; 27(7): 2287.

37.

Elkaeed

Alsfouk

Ibrahim

, et al. Computer-assisted drug discovery of potential natural inhibitors of the SARS-CoV-2 RNA-dependent RNA polymerase through a multi-phase in silico approach. Antiviral Therapy 2023; 28(5): 13596535231199838.

38.

Metwaly

Alesawy

Alsfouk

, et al. Computer-assisted drug discovery of potential African anti-SARS-CoV-2 natural products targeting the helicase protein. Nat Prod Commun 2024; 19(4): 1934578X241246738.

39.

Elkaeed

Khalifa

Alsfouk

, et al. The discovery of potential SARS-CoV-2 natural inhibitors among 4924 African metabolites targeting the papain-like protease: a multi-phase in silico approach. Metabolites 2022; 12(11): 1122.

40.

Metwaly

Elwan

El-Attar

AZMM

, et al. Structure-based virtual screening, docking, ADMET, molecular dynamics, and MM-PBSA calculations for the discovery of potential natural SARS-CoV-2 helicase inhibitors from traditional Chinese medicine. J Chem 2022; 2022: 7270094.

41.

Shin

Mukherjee

Grewe

, et al. Papain-like protease regulates SARS-CoV-2 viral spread and innate immunity. Nature 2020; 587(7835): 657–662.

42.

Klemm

Ebert

Calleja

, et al. Mechanism and inhibition of the papain-like protease, PLpro, of SARS-CoV-2. EMBO J 2020; 39(18): e106275.

43.

FDA-approved Drug Library. 2022, https://www.selleckchem.com/screening/fda-approved-drug-library.html (accessed 19 November 2021).

44.

Carvalho

, Trincão J and João Romão M. X-ray crystallography in drug discovery. Adv Drug Discov 2010: 31–56.

45.

Hassell

Bledsoe

, et al. Crystallization of protein–ligand complexes. Acta Crystal Sect D: Biol Crystal 2007; 63(1): 72–79.

46.

Vidal

Garcia-Serna

Mestres

. Ligand-based approaches to in silico pharmacology. Chemoinform Comput Chem Biol 2011: 489–502.

47.

Schneider

Tanrikulu

Schneider

. Self-organizing molecular fingerprints: a ligand-based view on drug-like chemical space and off-target prediction. Futur Med Chem 2009; 1: 213–218.

48.

Spackman

McKinnon

. Fingerprinting intermolecular interactions in molecular crystals. Crystengcomm 2002; 4(66): 378–392.

49.

Chu

Wang

, et al. In silico design of novel benzohydroxamate-based compounds as inhibitors of histone deacetylase 6 based on 3D-QSAR, molecular docking, and molecular dynamics simulations. New J Chem 2020; 44(48): 21201–21210.

50.

Ieritano

Campbell

Hopkins

. Predicting differential ion mobility behaviour in silico using machine learning. Analyst 2021; 146(15): 4737–4743.

51.

Taha

Ismail

Ali

, et al. Molecular hybridization conceded exceptionally potent quinolinyl-oxadiazole hybrids through phenyl linked thiosemicarbazide antileishmanial scaffolds: in silico validation and SAR studies. Bioorgan Chem 2017; 71: 192–200.

52.

Heikamp

Bajorath

. How do 2D fingerprints detect structurally diverse active compounds? Revealing compound subset-specific fingerprint features through systematic selection. J Chem Inform Model 2011; 51(9): 2254–2265.

53.

Opo

FADM

Rahman

Ahammad

, et al. Structure-based pharmacophore modeling, virtual screening, molecular docking, and ADMET approaches for identification of natural anti-cancer agents targeting XIAP protein. Sci Rep 2021; 11(1): 1–17.

54.

Duan

Dixon

Lowrie

, et al. Analysis and comparison of 2D fingerprints: insights into database screening performance using eight fingerprint methods. J Mol Graph Model 2010; 29(2): 157–170.

55.

Sastry

Lowrie

Dixon

, et al. Large-scale systematic analysis of 2D fingerprint methods and parameters to improve virtual screening enrichments. J Chem Inform Model 2010; 50(5): 771–784.

56.

Kogej

Engvist

Blomberg

, et al. Multifingerprint-based similarity searches for targeted class compound selection. J Chem Inform Model 2006; 46(3): 1201–1213.

57.

Willett

. Molecular similarity approaches in chemoinformatics: early history and literature status. In: Bienstock

Shanmugasundaram

Bajorath

(eds) Frontiers in molecular design and chemical information science-Herman Skolnik Award Symposium 2015: Jürgen Bajorath (ACS Symposium series). Oxford: Oxford University Press, pp. 67–89.

58.

Maggiora

Vogt

Stumpfe

, et al. Molecular similarity in medicinal chemistry: miniperspective. J Med Chem 2014; 57(8): 3186–3204.

59.

Altamash

Amhamed

Aparicio

, et al. Effect of hydrogen bond donors and acceptors on CO2 absorption by deep eutectic solvents. Processes 2020; 8(12): 1533.

60.

Wan

Tian

Wang

, et al. In silico studies of diarylpyridine derivatives as novel HIV-1 NNRTIs using docking-based 3D-QSAR, molecular dynamics, and pharmacophore modeling approaches. RSC Adv 2018; 8(71): 40529–40543.

61.

Turchi

Cai

Lian

. An evaluation of in-silico methods for predicting solute partition in multiphase complex fluids: a case study of octanol/water partition coefficient. Chem Eng Sci 2019; 197: 150–158.

62.

Sullivan

Enoch

Ezendam

, et al. An adverse outcome pathway for sensitization of the respiratory tract by low-molecular-weight chemicals: building evidence to support the utility of in vitro and in silico methods in a regulatory context. Appl Vitro Toxicol 2017; 3(3): 213–226.

63.

Escamilla-Gutiérrez

Ribas-Apaico

Córdova-Espinoza

, et al. In silico strategies for modeling RNA aptamers and predicting binding sites of their molecular targets. Nucleos Nucleot Nucl Acid 2021; 40: 798–807.

64.

Kaushik

Kumar

Bharadwaj

, et al. Ligand-based approach for in-silico drug designing. In: Kaushik

Kumar

Bharadwaj

, et al. (eds) Bioinformatics techniques for drug discovery. New York: Springer, 2018, pp. 11–19.

65.

Zhang

Ren

, et al. Development of an in silico prediction model for chemical-induced urinary tract toxicity by using naïve Bayes classifier. Mol Divers 2019; 23(2): 381–392.

66.

Cross

Thompson

Rai

, et al. Comparison of several molecular docking programs: pose prediction and virtual screening accuracy. J Chem Inform Model 2009; 49(6): 1455–1474.

67.

Plewczynski

Łaźniewski

Augustyniak

, et al. Can we trust docking results? Evaluation of seven commonly used programs on PDBbind database. Mol Inform 2011; 32(4): 742–755.

68.

Liu

Shi

Zhou

, et al. Molecular dynamics simulations and novel drug discovery. Expert Opini Drug Discover 2018; 13(1): 23–37.

69.

RPD. Bank. Protein Data Bank, 2020, https://www.rcsb.org/structure/4OW0

70.

Kim

Iyer

, et al. CHARMM-GUI: a web-based graphical user interface for CHARMM. J Comput Chem 2008; 29(11): 1859–1865.

71.

Brooks

MacKerell

, et al. CHARMM: the biomolecular simulation program. J Comput Chem 2009; 30(10): 1545–1614.

72.

Lee

Cheng

Swails

, et al. CHARMM-GUI input generator for NAMD, GROMACS, AMBER, OpenMM, and CHARMM/OpenMM simulations using the CHARMM36 additive force field. J Chem Theory Comput 2016; 12(1): 405–413.

73.

Best

Zhu

Shim

, et al. Optimization of the additive CHARMM all-atom protein force field targeting improved sampling of the backbone phi, psi and side-chain chi(1) and chi(2) dihedral angles. J Chem Theory Comput 2012; 8(9): 3257–3273.

74.

Phillips

Braun

Wang

, et al. Scalable molecular dynamics with NAMD. J Comput Chem 2005; 26(16): 1781–1802.

75.

Valdés-Tresanco

Valiente

, et al. gmx_MMPBSA: a new tool to perform end-state free energy calculations with GROMACS. J Chem Theory Comput 2021; 17(10): 6281–6291.

76.

Chen

Zeng

Wang

, et al. Decoding the identification mechanism of an SAM-III riboswitch on ligands through multiple independent Gaussian-accelerated molecular dynamics simulations. J Chem Inform Model 2022; 62(23): 6118–6132.

77.

Chen

Wang

Pang

, et al. Effect of mutations on binding of ligands to guanine riboswitch probed by free energy perturbation and molecular dynamics simulations. Nucleic Acid Res 2019; 47(13): 6618–6631.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB