Sage Journals: Discover world-class research

Abstract

Background

The COVID-19 pandemic has created an urgent need for effective therapeutic agents. The SARS-CoV-2 Main Protease (M^pro) plays a crucial role in viral replication and immune evasion, making it a key target for drug development. While several studies have explored M^pro inhibition, identifying FDA-approved drugs with potential efficacy remains a critical research focus.

Purpose

This study aims to identify FDA-approved drugs that could inhibit SARS-CoV-2 M^pro. Using computational screening, we seek compounds that share structural similarities with a known co-crystallized ligand (PRD_002214) and exhibit strong binding affinity to the enzyme, providing viable candidates for COVID-19 treatment.

Research Design

A systematic in silico approach was used, screening 3009 FDA-approved drugs. The initial screening focused on structural similarity to PRD_002214 (PDB ID: 6LU7), followed by molecular docking studies to predict binding affinity. Promising compounds were further analyzed through molecular dynamics (MD) simulations to evaluate their stability and interactions with M^pro over 100 ns.

Study Sample

Of the 3009 FDA-approved drugs screened, 74 were selected for initial evaluation. After refinement, 28 compounds underwent docking analysis, with eight showing strong binding potential to M^pro.

Analysis

Molecular docking assessed the binding affinity and interaction of the selected compounds with M^pro. MD simulations were conducted on the top compound, Atazanavir, to study its dynamic interactions. MM-GBSA, PLIP, and PCAT analyses were used to validate binding affinity and interactions.

Results

Eight compounds, including Carfilzomib, Atazanavir, Darunavir, and others, exhibited promising binding affinities. Among them, Atazanavir showed the highest binding strength and was selected for further MD simulation studies. These simulations revealed that Atazanavir forms stable interactions with M^pro, demonstrating favorable binding and dynamic stability. The binding affinity was further confirmed through MM-GBSA, PLIP, and PCAT analyses, supporting Atazanavir's potential as an effective M^pro inhibitor.

Conclusions

In silico results suggest that Atazanavir is a promising candidate for targeting SARS-CoV-2 M^pro, with strong binding affinity and dynamic stability. These findings support its potential as a lead compound for further preclinical and clinical testing, though in vitro and in vivo validation are needed to confirm its therapeutic efficacy against COVID-19.

Keywords

SARS-CoV-2 main protease drug repurposing FDA-approved drugs molecular docking molecular dynamics simulation Atazanavir

Introduction

As of May 2024, the World Health Organization reported 774 million confirmed cases of SARS-CoV-2 infection and nearly seven million fatalities, painting a sombre picture of the ongoing crisis. These alarming statistics underscore the immediate and pressing need for concerted efforts from the global scientific community to combat the virus and protect human health.

The usual drug discovery process is excessively expensive and time-consuming, typically taking 12 years to complete the discovery of a new drug, with average costs reaching 2.6 billion USD.¹ On the other hand, drug repurposing could be a very effective process in developing molecules or drugs indicated for new biological uses. It has the advantage of reducing the drug discovery timeline, as any existing compound already demonstrates a good safety profile in humans.^2,3 Consequently, there is no requirement for conducting Phase I clinical trials as part of the process.⁴ The drug repurposing strategy was applied successfully and severally in the revelation of anti-cancer,⁵ COVID-19,⁶ anti-inflammatory,⁷ antibacterial,⁸ antiparasitic,⁹ and antiviral¹⁰ drugs.

Fortunately, recent advancements in computational methodologies offer a promising tool for accelerating the drug discovery process.^11,12 In the past 20 years, In Silico virtual screening, a method employing computer algorithms to forecast the biological activity of compounds by analysing their chemical structures and their expected biological activities has surged in popularity. This includes rational drug design,^13,14 molecular docking,¹⁵ molecular dynamics (MD) simulations,¹⁶ pharmacophore studies,¹⁷ structure similarity assessments,¹⁸ molecular fingerprinting techniques,¹⁹ quantitative structure-activity relationship (QSAR) analyses,²⁰ homology modelling,²¹ as well as molecular modelling.²²

Against COVID-19, our research team utilized computational techniques in the pursuit of potential treatments. Specifically, we aim to screen a diverse array of natural compounds including a set of 310 natural antiviral metabolites,^23–27 4924 African natural metabolites,^28–30 5956 traditional Chinese medicine metabolites³¹ as well as 3009 FDA-approved drugs^32–35 for their ability to inhibit key SARS-CoV-2 proteins. By targeting these pivotal viral enzymes, we aim to find new treatments that can slow down the virus and lessen the strain on healthcare systems worldwide. Through interdisciplinary collaboration and innovative approaches, we are committed to making meaningful contributions to the ongoing battle against the COVID-19 pandemic.

In this work, a collection of 3009 FDA-approved drugs³⁶ has been screened using different computational methods to select the most potent SARS-CoV-2 main protease, M^pro, inhibitors. The utilized methods included molecular structures similarity and fingerprint studies against the co-crystallized ligand (PRD_002214) of SARS-CoV-2 M^pro PDB ID: 6LU7) (Figure 1), molecular docking, molecular dynamic (MD) simulations, MM-GPSA, PLIP, ProLIF and PCAT calculations against SARS-CoV-2 M^pro (PDB ID: 6LU7). This PDB model was selected due to its high resolution and extensive validation, making it a reliable reference for studies aimed at identifying potential inhibitors of M^pro. The presence of a co-crystallized ligand not only provides a benchmark for assessing the binding interactions of the examined compounds but also serves as a reference to elucidate structural similarity.

Figure 1.

Chemical structure of the co-crystallized ligand (PRD_002214) of SARS-CoV-2 M^pro, PDB ID: 6LU7.

Results and discussion

Filter using fingerprints

The investigation of molecular fingerprints was employed as an approach to analyse the relationship between compound structures and their biological effects. This methodology relies on Structure-Activity Relationship principles, indicating that compounds with akin structures tend to display similar biological effects.³⁷ In this study, a co-crystallized ligand (PRD_002214) exhibiting high affinity for M^pro (PDB ID: 6LU7), served as a benchmark drug. Compounds structurally akin to PRD_002214 are anticipated to also demonstrate strong binding and inhibition of the target M^pro enzyme. During the molecular fingerprints analysis, the computer derives physical and chemical descriptors from both reference and test compounds. These descriptors are encoded as mathematical symbols (bit strings), denoting the presence or absence of specific characteristics at each atom. Subsequently, these computed strings are compared to ascertain compound similarities.³⁸ We employed Discovery Studio software to evaluate the molecular fingerprints of 3009 FDA-approved drugs, juxtaposing them with PRD_002214, co-crystallized with SARS-CoV-2 M^pro. The objective was to identify the top compounds exhibiting the greatest resemblance to PRD_002214. The molecular fingerprints analysis encompassed the examination of various descriptors within the fragments and atoms of both test compounds and PRD_002214. These descriptors encompassed H-bond acceptor atom numbers,³⁹ H-bond donor atom numbers,⁴⁰ charge numbers,⁴¹ types of hybridization,⁴² number of positive ionizable atoms,⁴³ number of negative ionizable atoms,⁴⁴ number of halogens,⁴⁵ number of aromatic groups,⁴⁶ align with the ALogP.⁴⁷ As shown in Table 1, 74 compounds exhibited a high degree of similarities.

Table 1.

Fingerprint similarity analysis of 74 FDA-approved drugs with the highest similarity to PRD_002214 among 3009 FDA-tested compounds.

Comp.	Similarity	SA	SB	SC	Comp.	Similarity	SA	SB	SC
PRD_002214	1.000	1116	0	0	1673	0.827	1175	305	−59
30	0.781	1204	426	−88	1735	0.689	726	−63	390
31	0.708	1248	646	−132	1749	0.715	822	34	294
34	0.776	1077	272	39	1781	0.663	1124	579	−8
35	0.705	906	169	210	1787	0.724	1264	630	−148
90	0.677	770	21	346	1798	0.794	954	86	162
132	0.724	723	−117	393	1803	0.829	929	5	187
140	0.684	912	217	204	1808	0.682	708	−78	408
153	0.717	777	−33	339	1862	0.686	713	−76	403
156	0.683	688	−109	428	1892	0.677	700	−82	416
164	0.681	762	3	354	1910	0.679	958	295	158
220	0.679	745	−19	371	1931	0.687	739	−40	377
263	0.696	865	127	251	1959	0.691	699	−105	417
265	0.682	821	87	295	1972	0.668	674	−107	442
312	0.744	1037	277	79	1986	0.675	836	123	280
313	0.668	764	28	352	1991	0.671	827	116	289
314	0.671	644	−156	472	1997	0.713	1305	714	−189
333	0.690	691	−114	425	1998	0.727	1023	292	93
345	0.669	744	−4	372	2032	0.665	709	−50	407
368	0.730	878	87	238	2059	0.704	767	−26	349
372	0.698	880	145	236	2079	0.709	868	109	248
395	0.748	1254	560	−138	2184	0.726	802	−11	314
401	0.698	937	226	179	2187	0.677	815	88	301
402	0.754	1036	258	80	2190	0.664	914	260	202
420	0.665	934	288	182	2207	0.682	840	115	276
424	0.666	733	−16	383	2213	0.721	1538	1017	−422
443	0.692	895	177	221	2216	0.792	1092	263	24
532	0.687	766	−1	350	2217	0.721	1465	917	−349
582	0.678	795	56	321	2218	0.741	1445	835	−329
606	0.666	763	29	353	2221	0.690	1751	1420	−635
609	0.696	726	−73	390	2222	0.709	1658	1224	−542
611	0.704	748	−53	368	2224	0.767	1546	899	−430
631	0.667	757	19	359	2225	0.712	1539	1047	−423
684	0.723	708	−137	408	2229	0.714	822	36	294
697	0.854	1155	236	−39	2240	0.663	680	−91	436
712	0.696	865	127	251	2269	0.683	772	15	344
774	0.696	679	−141	437	2273	0.674	743	−14	373
811	0.710	772	−29	344	2305	0.693	1391	891	−275
820	0.710	975	257	141	2320	0.671	1046	444	70
945	0.716	969	237	147	2323	0.684	962	290	154
947	0.686	831	95	285	2324	0.746	952	160	164
976	0.667	733	−17	383	2387	0.670	860	168	256
986	0.674	698	−80	418	2413	0.791	933	63	183
1010	0.705	1306	737	−190	2440	0.737	974	206	142
1011	0.727	1119	423	−3	2468	0.766	1113	337	3
1012	0.713	1235	615	−119	2486	0.741	879	71	237
1021	0.690	686	−122	430	2497	0.774	846	−23	270
1029	0.747	1110	369	6	2527	0.767	844	−15	272
1032	0.686	1394	916	−278	2536	0.726	970	220	146
1034	0.731	1205	532	−89	2548	0.696	906	185	210
1037	0.784	890	19	226	2669	0.683	966	298	150
1059	0.685	736	−41	380	2689	0.686	673	−135	443
1083	0.729	1192	518	−76	2707	0.703	754	−43	362
1183	0.676	973	324	143	2711	0.667	790	68	326
1189	0.671	934	275	182	2712	0.684	802	57	314
1200	0.745	974	192	142	2718	0.759	928	107	188
1233	0.692	922	217	194	2728	0.675	802	73	314
1259	0.816	960	61	156	2756	0.738	888	87	228
1270	0.774	1105	312	11	2765	0.738	1096	369	20
1293	0.676	1823	1582	−707	2777	0.714	835	53	281
1309	0.751	892	71	224	2787	0.668	1531	1175	−415
1333	0.679	1073	465	43	2818	0.668	1461	1071	−345
1340	0.740	878	71	238	2856	0.691	935	238	181
1397	0.722	953	204	163	2864	0.674	852	149	264
1433	0.667	736	−13	380	2870	0.668	736	−15	380
1448	0.807	1003	127	113	2873	0.693	747	−38	369
1450	0.768	1004	192	112	2891	0.705	835	69	281
1452	0.674	749	−5	367	2902	0.667	723	−32	393
1457	0.690	970	290	146	2905	0.715	1200	562	−84
1526	0.679	934	259	182	2922	0.696	923	211	193
1531	0.722	806	1	310	2924	0.754	1051	277	65
1552	0.717	884	117	232	2934	0.679	1222	685	−106
1600	0.667	731	−20	385	2961	0.693	710	−92	406
1657	0.674	1006	377	110	2971	0.667	834	134	282
1666	0.747	1002	226	114	2978	0.719	1482	944	−366

SA: The number of bits in both PRD_002214 and the examined drug.

SB: The number of bits in the examined drug but not the PRD_002214.

SC: The number of bits in PRD_002214 but not the examined drug.

Molecular similarity

Although the molecular similarity study shares the in Silico nature and ligand comparison foundation with the molecular fingerprints study, it differs in its methodology. This study involves the comprehensive description and comparison of the entire structures of both the reference compound and the compounds under examination. It employs descriptors of steric, topological, electronic, and physical properties, contrasting with the molecular fingerprints study, which concentrates on atomic descriptors within fragments, atoms, or substructures.⁴⁸ To execute the molecular similarity study, 74 FDA-approved drugs demonstrating the highest similarity based on the molecular fingerprints study were chosen. These drugs were compared to PRD_002214 using Discovery Studio software. The investigation encompassed the scrutiny of various descriptors (illustrated in Figure 2 and Table 2), including partition coefficient (ALog p), molecular weight (M-W), number of H-bond donor atoms (HB-A), number of H-bond acceptor atoms (HB-D), number of rotatable bonds (R-B), number of rings (R) and aromatic rings (A-R), minimum distance, and molecular fractional polar surface area (MFPSA).⁴⁹ The objective was to pinpoint the most akin candidates among the examined FDA-approved drugs. As shown in Figure 3, 28 compounds exhibited the highest degree of similarities.

Figure 2.

The results of the molecular similarity analysis conducted between PRD_002214 (depicted as a green ball) and the 74 most closely related FDA-approved drugs. It visually represents similar drugs (depicted in green) and those that are dissimilar (depicted as blue balls).

Table 2.

The names, codes, and structural features of 28 FDA-approved drugs that exhibit similarities with PRD_002214.

Comp.	ALog p	M. Wt	HB-A	HB-D	R-B	R	A-R	MFPSA	Minimum distance
PRD_002214	2.453	680.791	8	5	18	3	2	0.273	0.000
35	4.978	720.944	7	4	18	4	4	0.264	0.687
140	4.926	628.801	5	4	15	4	3	0.178	0.681
333	−0.502	492.519	10	4	12	2	1	0.325	0.754
402	4.083	802.934	13	7	18	3	3	0.29	0.656
532	2.747	634.606	11	4	9	4	2	0.38	0.774
697	3.799	719.91	8	4	20	4	2	0.203	0.403
774	1.266	508.634	7	6	10	3	1	0.361	0.755
811	2.622	593.732	9	4	12	4	2	0.275	0.460
1021	0.921	529.453	9	3	11	3	1	0.315	0.662
1029	2.685	679.849	8	4	14	5	1	0.252	0.553
1037	4.871	663.888	9	5	10	4	2	0.272	0.591
1059	2.714	527.649	8	3	13	2	2	0.208	0.556
1200	4.69	704.855	9	5	18	3	3	0.216	0.453
1448	3.565	748.286	10	3	14	5	2	0.252	0.561
1457	1.366	609.814	10	3	12	3	0	0.242	0.760
1552	4.129	585.644	7	0	15	3	1	0.283	0.638
1673	3.46	717.979	8	4	20	2	1	0.179	0.543
1735	−0.137	527.604	11	4	18	1	1	0.282	0.699
1892	4.269	500.548	6	3	14	2	1	0.187	0.718
2032	2.631	547.664	8	3	12	4	2	0.274	0.514
2059	−0.343	588.716	10	4	11	4	2	0.309	0.635
2079	−2.255	652.816	12	6	21	2	2	0.232	0.723
2387	1.356	702.888	14	2	19	2	2	0.315	0.695
2497	3.327	551.758	7	4	19	1	1	0.222	0.601
2536	6.536	680.935	10	6	14	4	2	0.218	0.625
2548	2.475	711.868	11	6	12	5	3	0.272	0.674
2669	5.754	656.723	13	1	21	3	1	0.258	0.785
2991	2.174	602.576	12	4	14	4	3	0.356	0.676

Figure 3.

Twenty-eight compounds with good molecular similarity with the co-crystallized ligand (PRD_002214) of COVID-19 main protease (PDB ID: 6LU7).

Docking studies

In the current study, docking was inescapable as it is the only computational method capable of predicting the exact binding between the examined members and COVID-19 main protease (PDB ID: 6LU7). However, the docking process is prone to error and requires proper validation. So, the co-crystallized ligand (PRD_002214) was re-docked to COVID-19 main protease. The calculated RMSD between docked and co-crystalized poses was 0.48, indicating a valid docking approach.

PRD_002214 achieved an energy score of −30.25 Kcal/mol. It showed eight hydrogen bonds and four hydrophobic interactions. The 2-oxopyrrolidin-3-yl moiety occupied the first pocket of M^pro forming two hydrogen bonds with His163 and Glu166. The 2-acetamido-3-methylbutanamido)-N-ethyl-4-methylpentanamide moiety occupied the second pocket of M^pro forming four hydrogen bonds with Gln189, Glu166, and Thr190 besides three hydrophobic interactions with Met165 and His41. Furthermore, the benzyl acetate moiety occupied the third pocket of the receptor forming two hydrogen bonds with His164 and His41. Moreover, 5-methylisoxazole-3-carboxamide moiety was incorporated in the fourth pocket forming a hydrophobic interaction with Ala191 (Figure 4).

Figure 4.

(A) 3D, (B) 2D of (PRD_002214) in the active site of COVID-19 main protease. (C) 3D binding mode of compound 697 into 6lu7 active site. (D) 2D binding mode of compound 697 in the 6lu7 active site.

Based on the results obtained from docking studies (Table 3), eight compounds (carfilzomib 697, Atazanavir 1200, Darunavir 811, Monomethyl auristatin E 1673, Telaprevir 1029, Nelfinavir 1037, Aliskiren 2497, and Fosinopril 1552), among the 28 filtered members, were able to achieve better docking scores (−33.59, −33.06, −31.33, −33.70, −32.64, −31.93, −32.29, and −31.22 Kcal/mol, respectively) than PRD_002214 (−30.25 Kcal/mol), and thus were selected for further analysis.

Table 3.

The calculated ∆G (binding free energies) of the tested compounds and co-crystallized ligand against COVID-19 main protease (∆G in Kcal/mol).

Compound no.	Compound name	∆G in (Kcal/mol)
PRD_002214	The co-crystallized ligand	−29.27
35	Ritonavir	−30.03
140	Lopinavir	−28.8
333	Enalapril	−27.07
532	Novobiocin	−20.49
697	Carfilzomib	−33.59
774	Argatroban	−26.88
811	Darunavir	−31.33
1021	Sofosbuvir	−27.11
1029	Telaprevir	−32.64
1037	Nelfinavir	−31.93
1059	Fesoterodine	−31.05
1200	Atazanavir	−33.06
1448	Asunaprevir	−29.02
1457	Tiamulin	−29.63
1552	Fosinopril	−31.22
1673	Monomethyl auristatin E (MMAE)	−33.7
1735	Oxeladin	−26.57
1892	Travoprost	−28.59
2032	Darunavir	−29.57
2059	Penicillin G Procaine	−20.56
2079	Metroprolol	−21.85
2387	Sulbutiamine	−27.64
2497	Aliskiren	−32.29
2536	(S)-Penbutolol	−26.8
2548	Indinavir	−29.78
2669	Riboflavin	−27.73
2991	Remdesivir	−27.55

As shown in Figure 5, compound 697 was able to engage in many diverse interaction types with the M^pro. It was involved in three hydrogen-bond interactions with residues Gly143, Ser144, and Cys145 and one pi-pi interaction with His163 through its oxopentan moiety. In addition, the 4-phenylbutanamido moiety achieved two hydrophobic interactions with Pro168 and Met49 besides two H bonds with Gln189 and Glu166.

Figure 5.

(A) 3D binding mode of compound 1200 into 6lu7 active site. (B) 2D binding mode of compound 1200 in the 6lu7 active site.

A closer look at compound 1200 shows its engagement in a number of diverse interaction types within the M^pro binding site. Compound 1200 was able to make two electrostatic interactions with His164 and Asp187 besides two pi-pi interactions with Met165 and Cys85 via its pyridine moiety. Also, the benzyl moiety closed to the pyridine ring was involved in one electrostatic and one pi-pi interactions with Met165 and His41, respectively. Moreover, Compound 1200 was buried in the main different pockets of the M^pro through the formation of four H-bonding interactions with residues Glu166, Gln189, Asn142, and Gly143 as described in Figure 5.

The proposed binding mode of 1673 displayed that it occupied the four different pockets of the main protease through the formation of three hydrogen bonds and three hydrophobic interactions. The 1-hydroxy-1-phenyl moiety formed one H-bond with Glu166 while 1-methoxy-2-methyl-3-oxopropyl moiety was involved in one H-bonding interaction with Gln189 and two pi-pi interactions with Met49 and His41. Also, pyrrolidin moiety was buried inside the M^pro active pocket to form one pi-pi interaction with Cys145. Finally, the oxoheptan moiety formed another H-bond with Asn142 through its oxo group as depicted in Figure 6.

Figure 6.

(A) 3D binding mode of compound 1673 into 6lu7 active site. (B) 2D binding mode of compound 1673 in the 6lu7 active site.

Respecting the docking results of 1029 (affinity value of −32.64 kcal/mol), it revealed the formation of many H-bonding and hydrophobic interactions in the different pockets of the main protease active pocket. At first, the pyrazine-2-carboxamido moiety formed one hydrogen bond with Glu166, one pi-pi interaction with Pro168, and one electrostatic interaction with Met165. Likely, the ter-butyl moiety formed three hydrophobic interactions with Met165, His41, and His164. Moreover, compound 1029 was buried in the active pocket through the formation of a hydrogen bond with Cys145 and five hydrophobic interactions with Met49, Pro52, Cys44, and His41using its octahydrocyclopenta[c]pyrrole moiety (Figure 7).

Figure 7.

(A) 3D binding mode of compound 1029 into 6lu7 active site. (B) 2D binding mode of compound 1029 in the 6lu7 active site.

MD simulations

Both the distance between Atazanavir’s centre of mass and the centre of mass of the M^pro and the structure of the holo M^pro become stable either from the start of the simulation or after 20 ns. The apo M^pro protein (blue) shows a stable RMSD average with a value of 2.1 Å during the first 40 nanoseconds. After that, the RMSD values start to fluctuate and increase in two steps. From 45 ns to around 80 ns, it shows an average of 3.3 Å then it increases again to an average of around 5 Å during the last 15 ns. On the other hand, the holo M^pro protein (red) displays a constant trend after the first 20 ns of the simulation, keeping a value of around 3 Å on average (Figure 8(A)). The RMSD of Atazanavir shows two states during the simulation. After having an average of 5.8 Å for the first 25 nanoseconds, it jumps up to a maximum of 14 Å then it fluctuates from 30 to 45 ns before stabilizing around an average of 9.6 Å (Figure 8(B1)). These changes in the RMSD values result from a conformational adjustment of Atazanavir within the binding pocket, indicating a dynamic binding process before stabilization. As seen in Figure 8(B2), a comparison of Atazanavir’s conformation at 6 ns (green sticks) with those at 40 ns (cyan sticks) and 65 ns (magenta sticks) reveals notable positional shifts, reflecting its structural accommodation within the M^pro active site. This reorientation likely allows Atazanavir to optimize its interactions, enhancing binding stability over time. These adjustments in binding orientation highlight Atazanavir’s flexibility and suggest a gradual establishment of a stable, energetically favourable conformation by the end of the simulation.

Figure 8.

MD simulations analyses related to the Atazanavir-M^pro complex trajectory. (A) RMSD values from the trajectory for the M^pro protein, (B1) The Atazanavir’s RMSD values, (B2) A comparison between the Atazanavir conformation at 6 ns (green sticks) with 40 ns (cyan sticks) and 6 ns with 65 ns (magenta sticks), (C) radius of gyration for the M^pro protein, (D) SASA for the M^pro protein, (E) change in the number of hydrogen bonds for the M^pro protein, (F) RMSF for the M^pro protein, (G) distance from the centre of mass of Atazanavir and M^pro protein. Blue and red lines represent the M^pro protein in apo and holo forms, respectively.

As can be seen in Figure 8(C) (RoG), and Figure 8(D) (SASA), the apo M^pro system and the holo M^pro system have nearly the same average during the first 40 ns. After that, the apo M^pro protein shows a slight increase in both the RoG (around 0.5 Å increase) and SASA (around 500 Å²) values. Figure 8(E) indicates that the apo and holo M^pro systems have a virtually similar amount of H-bonds on average (70 bonds). In addition, the oscillations of the C-alpha atoms in the apo M^pro protein show nearly always a higher value than the corresponding RMSF of the holo M^pro protein (Figure 8(F)). In general, this suggests that the binding of Atazanavir may have a stabilizing effect on the M^pro protein structure. Interstingly, Atazanavir maintains a consistent average separation from the M^pro protein’s centre of mass, approximately 22.6 Å, which indicates a relatively stable interaction throughout the simulation (Figure 8(G)). This steady distance suggests that, despite conformational adjustments within the binding site, Atazanavir does not drift significantly from its initial binding region. Such stability in spatial positioning relative to the protein’s centre of mass supports the idea of a sustained, reliable binding mode, which could contribute to its effectiveness as an inhibitor by maintaining close and continuous interactions with the active site residues of M^pro.

Molecular mechanics-generalized born surface area (MM-GBSA) calculations

MM-GBSA technique was utilized to estimate the binding free energies of the Atazanavir- M^pro complex. MM-GBSA combines molecular mechanics energy terms, solvation-free energies, and surface area contributions to assess the overall stability and affinity of the examined Atazanavir-M^pro interaction. It is commonly employed in drug discovery and computational biology to predict and optimize ligand binding affinities.⁵⁰ Figure 9 shows the various components of the predicted binding free energy estimated using the MM-GBSA technique. A moderate interaction may be derived from the fact that the binding energy of the Atazanavir molecule is −26.16 kcal/mol, indicating a moderate interaction between the two entities. Van der Waals interactions seem to be more important than electrostatic interactions in influencing binding stability (−37.65 Kcal/Mol versus −11 Kcal/Mol). Using decomposition analysis, we were able to measure the contributions of amino acids that are within 1 nm of the Atazanavir molecule (Figure 10). Met49 (−1.92 Kcal/Mol), Met165 (−1.41 Kcal/Mol), and Gln189 (−3.15 Kcal/Mol) are amino acids that are within 1 nm and contribute more than −1 Kcal/Mol.

Figure 9.

Various energetic components of MM-GBSA along with their respective values, with bars representing the standard deviations.

Figure 10.

Binding free energy decomposition of the M^pro-Atazanavir complex.

Protein-ligand interaction fingerprints (ProLIF) calculations

ProLIF is a computational method that used in this research as a molecular modelling tool to analyse and characterize the M^pro-Atazanavir complex interactions. ProLIF generated fingerprints or descriptors that encoded information about the types and spatial arrangements of interactions between the M^pro-Atazanavir complex. These fingerprints provide valuable insights into the binding mode, affinity, and selectivity of the Atazanavir complex, aiding in the understanding of interaction and binding. According to the ProLIF library results, five amino acids were discovered to have a higher than 71% incidence of hydrophobic interaction. His41 (78.9 %), Met49 (94.6 %), Met165 (94.7 %), Pro168 (71.1 %), and Gln189 (96.6 %) are the most prevalent amino acids that interact with the Atazanavir molecule. Figure 11 illustrates the amino acids involved, the nature of interactions with Atazanavir, and their frequency throughout the entire simulation period, as analysed using the ProLIF Python library.

Figure 11.

The amino acids involved, the nature of interactions with the M^pro-Atazanavir complex.

Protein-ligand interaction profiler (PLIP) calculations

PLIP was used to identify and characterize different types of interactions between M^pro and Atazanavir, such as hydrogen bonds, hydrophobic contacts, pi-stacking interactions, and salt bridges. PLIP helps researchers to understand the structural basis of ligand binding and to predict the binding affinity and specificity of protein-ligand complexes. As Figure 12 shows, four different clusters were obtained by the PLIP exhibiting the 3D binding interactions as . pse files from the representative frames obtained from TTClust and their 3D interactions (hydrophobic interactions and H-bonds) with the M^pro-Atazanavir complex.

Figure 12.

The four representative clusters derived from TTClust and their three-dimensional interactions with the M^pro-Atazanavir complex. Hydrophobic interactions are represented by grey dashed lines, hydrogen bonds by blue solid lines, Atazanavir by orange sticks, and amino acids of the M^pro protein by blue sticks.

Principal component analysis trajectory (PCAT) calculations

PCAT was utilized to analyse and visualize the collective motions of atoms in the M^pro-Atazanavir complex. PCAT reduces the dimensionality of the trajectory data by identifying the principal components of motion, which are the major modes of structural variation in the M^pro-Atazanavir complex. By projecting the trajectory onto these principal components, PCAT enables the identification of key conformational changes, transitions, and dynamic behaviours of the M^pro-Atazanavir complex. The size of the reduced subspace was determined using the scree plot, the distribution of the eigenvectors, and the variance maintained with new eigenvectors (cumulative sum), as stated in the methods section. The slope seems to level off around the second PC as shown in the scree plot. The first eigenvector alone accounted for around 63 % of the total variance, whereas the sum of the first three eigenvectors accounted for over 75% of the whole variance (Figure 13). It was shown that the distribution of the first three PCs did not follow a Gaussian pattern (Figure 14). Therefore, we chose the top three eigenvectors to represent the reduced subspace.

Figure 13.

The variation in eigenvalues as the number of eigenvectors increases (depicted by the blue line). Additionally, the cumulative variance preserved by these eigenvectors (represented by the red line).

Figure 14.

The distribution of the first 10 eigenvectors.

The cosine content was determined for both the apo and holo M^pro simulations, allowing an assessment of the degree of randomness shown by the behaviour of the first 10 eigenvectors. Except for the first principal component of the apo protein, which attained a value of 0.23, the cosine content of the first 10 eigenvectors in both the apo and holo proteins was less than 0.2. (Figure 15). The Root Mean Square Inner Product (RMSIP) shows that the similarity between the two subspaces (the first three eigenvectors) is 26.3 %, indicating that there is a considerable difference between the two trajectories. Furthermore, the RMSIP found that the C matrices of apo and holo proteins are only 34.4 % comparable, suggesting that the sampling was different in both systems.

Figure 15.

The cosine content values of the first 10 eigenvectors for both trajectories.

Figure 16 illustrates the results of projecting each trajectory onto the first three eigenvectors of the new C matrix. In each of these graphs, the larger dot represents the average structure of the corresponding trajectory. Figure 16(A), a projection on the first two eigenvectors, shows that the two trajectories have nearly identical average structures and that the frames reflect distinctive sampling with just a little degree of overlap at the start of the simulation (pale red and white dots). Furthermore, as compared to the apo protein, the sampling of the holo protein trajectory suggests more similar frames for all of the simulation time. Figure 16(B) shows that there is a minor overlap between the two trajectories and that their average structures are comparable. Furthermore, throughout the simulation, the frames of the holo system (shown by red dots) continued to exhibit a clustering tendency similar to that exhibited in Figure 9. The apo protein, on the other hand, has a high degree of variation between its frames. Similarly, the projection on the second and third eigenvectors, shown in Figure 16(C), follows the same pattern as the previous projections, with the apo protein exhibiting a considerable disparity between its frames and the holo protein exhibiting closely clustered frames.

Figure 16.

The projections of each trajectory; (A) the first two, (B) the first and third, (C) the second and third eigenvectors.

Porcupine diagrams, employed in Figure 17(A) and (C), provide a detailed depiction of the dynamics dictated by the first three eigenvectors. In these diagrams, amino acids crucial to the binding pocket are distinguished by a blue colouration. Particularly noteworthy is the observation that within the apo M^pro structure (depicted in a green cartoon representation), these blue amino acids undergo substantial motion, ultimately leading to the closure of the binding pocket evident in both the first and second principal components (PCs) as illustrated in Figure 17(A) and (B). Similarly, while examining the holo protein (depicted in a red cartoon representation), it becomes apparent that its amino acids undergo similar motions, albeit with diminished magnitudes. Interestingly, the analysis of the third PC (Figure 17(C)) unveils a distinct behaviour: while the motion of the apo protein tends to close the pocket, the holo protein’s motion prompts the pocket to open, suggesting a nuanced interplay between protein dynamics and ligand binding states.

Figure 17.

The porcupine figures of each of the first three eigenvectors (A) shows the first PC, (B) shows the second PC, and (C) shows the third PC) for both systems. Green cartoon: apo M^pro trajectory, red cartoon: holo M^pro trajectory. Blue cartoon: amino acids lining the binding pocket.

While our computational findings provide promising insights into the binding stability and potential inhibitory activity of Atazanavir against SARS-CoV-2 M^pro, experimental validation is essential. Future in vitro and in vivo studies will be necessary to confirm the efficacy, specificity, and safety of these inhibitors under physiological conditions. Such studies will provide a deeper understanding of pharmacokinetics, potential off-target effects, and therapeutic viability, ultimately advancing Atazanavir towards clinical application.

However, Atazanavir has shown promise as a SARS-CoV-2 inhibitor, aligning with studies that support its potential for repurposing in COVID-19 treatment. Previous research demonstrates that Atazanavir effectively inhibited the viral replication in both cell-based assays and animal models.⁵¹ In combination with ritonavir, Atazanavir not only reduces SARS-CoV-2 replication in vitro but also diminishes virus-induced inflammatory responses, specifically interleukin-6 and TNF-α levels, in human lung cell lines.⁵² Additionally, recent work using mouse models has shown that Atazanavir provides partial protection against SARS-CoV-2 challenge, suggesting its activity may extend across different viral variants.⁵³ These findings strengthen our study’s implications by situating Atazanavir as a promising M^pro inhibitor, meriting further exploration in clinical settings to fully validate its antiviral efficacy against SARS-CoV-2.

Materials and method

Molecular fingerprint studies

Discovery Studio 4.0 software was employed to analyse the molecular fingerprint between 3009 FDA-approved drugs and PRD_002214, as previously outlined.⁵⁴ Default atomic parameters, including the number of charges, halogen atoms, aromatic rings, hydrogen-bond acceptors and donors, as well as positive and negative ionizable atoms, were determined. Additionally, hybridization and ALogP were assessed to identify compounds with the closest resemblance. The methodology section in the Supplemental data offers a comprehensive explanation of the applied approach, providing detailed descriptions of the procedures utilized.

Structural similarity studies

Discovery Studio 4.0 software was utilized to analyse the molecular similarities between 74 selected FDA-approved drugs and PRD_002214, as detailed previously.^55,56 Default molecular properties, including the number of rotatable bonds, rings, aromatic rings, hydrogen-bond donors and acceptors, along with ALog p, molecular weight, and MFPSA, were employed to identify the most similar compounds. The methodology section in the Supplemental data offers a comprehensive explanation of the applied approach, providing detailed descriptions of the procedures utilized.

Docking studies

Using the Discovery Studio software,^57,58 docking studies against the SARS-CoV-2 M^pro (PDB ID: 6LU7) were conducted, with the co-crystallized ligand (PRD_002214) serving as a reference drug to guide the investigation. The methodology section in the Supplemental data provides a detailed description of the approach employed, outlining the procedures in depth. During the docking process, PRD_002214 was virtually docked into the Mpro active site protein, allowing assessment of its binding affinity and orientation. Utilizing a known and experimentally verified ligand as a reference provided insights into the potential interactions and binding modes of the most similar 28 compounds selected from the similarity study. The methodology section in the Supplemental data offers a comprehensive explanation of the applied approach, providing detailed descriptions of the procedures utilized.

Molecular dynamics simulations

To analyse the M^pro-Atazanavir complex’s stability and structural changes between the apo and holo forms, we conducted a 100 ns unbiased MD simulation in GROMACS 2021.^59–61 Input files were prepared using the CHARMM-GUI server,⁶² solvating the docked complex and apo protein in a 9.7 nm cube with TIP3P water and 0.154 M NaCl ions. Equilibration was performed to reach 310 K and 1 atm using the NVT and NPT ensembles. During production, temperature and pressure were maintained, and trajectory data was collected at 0.1 ns intervals.⁶³ Analysis involved RMSD calculations, hydrogen-bond measurements, and interaction analysis. The methodology section in the Supplemental data offers a comprehensive explanation of the applied approach, providing detailed descriptions of the procedures utilized.

MM-GBSA calculations

We calculated the ligand’s binding energy of the M^pro-Atazanavir complex using the MM-GBSA method with the gmx_MMPBSA program.⁶⁴ Additionally, decomposition analysis was conducted to determine the contribution of amino acids interacting within 1 nm of Atazanavir. The ionic strength was maintained at 0.154 M, and the solvation method (igb) was set to 5. Internal and external dielectric constants were set to 1.0 and 78.5, respectively. The methodology section in the Supplemental data offers a comprehensive explanation of the applied approach, providing detailed descriptions of the procedures utilized.

ProLIF calculations

The trajectory of the M^pro -Atazanavir complex underwent clustering via TTclust, a method used to group similar conformations within a molecular dynamics trajectory. This process identifies distinct conformational states by analysing the structural similarity of frames throughout the simulation. By clustering the trajectory, representative frames were extracted for each cluster, providing snapshots that encapsulate the diverse structural dynamics and conformational transitions of the M^pro -Atazanavir complex.⁶⁵ The methodology section in the Supplemental data offers a comprehensive explanation of the applied approach, providing detailed descriptions of the procedures utilized.

PLIP calculations

The PLIP was utilized to analyse these frames for the quantity and types of the M^pro-Atazanavir interactions. The resulting data was saved in a .pse file format, which can be visualized using PyMol.⁶⁶ The methodology section in the Supplemental data offers a comprehensive explanation of the applied approach, providing detailed descriptions of the procedures utilized.

PCAT calculations

PCA of MD trajectory atom mass-weighted covariance matrix (C) unveils system motion coordination. We focused on alpha carbon mobility in the M^pro protein, excluding terminal amino acids from Phe8 to Cys300. Alignment to the last frame of equilibrium stages served as the reference frame. PCA identifies eigenvectors capturing atomic motions, with eigenvalues indicating motion magnitude decreasing with eigenvector index. GROMACS tools, gmx covar and gmx anaeig, were used for C matrix diagonalization and analysis.⁶⁷ Determination of essential subspace principal components involved cumulative sum of eigenvalues, scree plot analysis, and examination of eigenvector distribution. The methodology section in the Supplemental data offers a comprehensive explanation of the applied approach, providing detailed descriptions of the procedures utilized.

Conclusion

Our comprehensive in Silico screening and molecular dynamics simulations have yielded promising results in identifying potential candidates for repurposing FDA-approved drugs to target the Main Protease of SARS-CoV-2. Among the compounds investigated, Carfilzomib, Atazanavir, Darunavir, Monomethyl auristatin E, Telaprevir, Nelfinavir, Aliskiren, and Fosinopril exhibited robust binding affinity and favourable interactions, indicating their potential as therapeutic agents against COVID-19. Atazanavir emerged as a particularly promising candidate, showing strong binding affinity and stability in MD simulations, along with positive interactions in MM-GBSA, PLIP, and PCAT analyses. These findings align with previously reported in vitro activities of Atazanavir and further provide new insights into its molecular-level activity, enhancing our understanding of its inhibitory potential.

Supplemental Material

Supplemental Material - Repurposing FDA-approved drugs for COVID-19: targeting the main protease through multi-phase in silico approach

Supplemental Material for Repurposing FDA-approved drugs for COVID-19: targeting the main protease through multi-phase in silico approach by Ahmed M Metwaly, Eslam B Elkaeed, Aisha A Alsfouk, Ibrahim M Ibrahim, Hazem Elkady and Ibrahim H Eissa in Antiviral Therapy.

Footnotes

Acknowledgements

The authors would like to thank AlMaarefa University, Riyadh, Saudi Arabia, for supporting this research.

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 iDs

Eslam B Elkaeed

Hazem Elkady

Supplemental Material

Supplemental material for this article is available online.

References

Chan

Shan

Dahoun

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

Parvathaneni

Kulkarni

Muth

, et al. Drug repurposing: a promising tool to accelerate the drug discovery process. Drug Discov 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 Discov Today Ther Strat 2011; 8(3-4): 61–69.

Sleire

Førde

Netland

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

Singh

Parida

Lingaraju

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

Hong

Bang

. Anti-inflammatory strategies for schizophrenia: a review of evidence for therapeutic applications and drug repurposing. Clin Psychopharmacol Neurosci 2020; 18(1): 10–24.

Konreddy

Rani

Lee

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

Shirley

D-A

Sharma

Warren

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

10.

Trivedi

Mohan

Byrareddy

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

11.

Hagler

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

12.

Engel

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

13.

Jalmakhanbetova

Elkaeed

Eissa

, et al. Synthesis and molecular docking of some grossgemin amino derivatives as tubulin inhibitors targeting colchicine binding site. J Chem 2021; 2021: 1–10.

14.

El-Adl

El-Helby

AGA

Ayyad

, et al. Design, synthesis, and anti-proliferative evaluation of new quinazolin-4 (3H)-ones as potential VEGFR-2 inhibitors. Bioorg Med Chem 2021; 29: 115872.

15.

Imieje

, et al. Antileishmanial derivatives of humulene from Asteriscus hierochunticus with in silico tubulin inhibition potential. Record Nat Prod 2021; 16: 150–171.

16.

Rafi

Al-Khafaji

Tok

, et al. Computer-based identification of potential compounds from Salviae miltiorrhizae against Neirisaral adhesion A regulatory protein. J Biomol Struct Dyn 2020; 40: 4301–4313.

17.

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 Dyn 2021; 39(15): 5638–5656.

18.

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.

19.

Baidya

Ghosh

Amin

, et al. In silico modelling, 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.

20.

Shi

. Support vector regression-based QSAR models for prediction of antioxidant activity of phenolic compounds. Sci Rep 2021; 11(1): 8806–8809.

21.

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

22.

Eissa

Ibrahim

Metwaly

, et al. Design, molecular docking, in vitro, and in vivo studies of new quinazolin-4 (3H)-ones as VEGFR-2 inhibitors with potential activity against hepatocellular carcinoma. Bioorg Chem 2021; 107: 104532.

23.

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.

24.

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.

25.

Eissa

Khalifa

Elkaeed

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

26.

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.

27.

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.

28.

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. Antivir Ther 2023; 28(5): 13596535231199838.

29.

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.

30.

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.

31.

Metwaly

Elwan

El-Attar

AAMM

, 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 the traditional Chinese medicine. J Chem 2022; 2022: 1–23.

32.

Metwaly

Elkaeed

Alsfouk

, et al. Repurposing FDA-approved drugs as potential inhibitors of SARS-CoV-2 PLpro: a comprehensive computational study. J Comput Bio Chem. 23: 1–23.

33.

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. Processes 2022; 10(3): 530.

34.

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: 1–19.

35.

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. Molecules 2022; 27(7): 2287.

36.

FDA-Approved drug library. 2022. https://www.selleckchem.com/screening/fda-approved-drug-library.html.

37.

Tong

Welsh

Shi

, et al. Structure-activity relationship approaches and applications. Environ Toxicol Chem 2003; 22(8): 1680–1695.

38.

Hassell

Bledsoe

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

39.

Spackman

McKinnon

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

40.

Chu

Q-X

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.

41.

Ieritano

Campbell

Hopkins

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

42.

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. Bioorg Chem 2017; 71: 192–200.

43.

Heikamp

Bajorath

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

44.

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): 4049.

45.

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.

46.

Sastry

Lowrie

Dixon

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

47.

Kogej

Engkvist

Blomberg

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

48.

Maggiora

Vogt

Stumpfe

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

49.

Bajorath

. Molecular similarity concepts for informatics applications. Methods Mol Biol 2017; 1526: 231–245.

50.

Genheden

Ryde

. The MM/PBSA and MM/GBSA methods to estimate ligand-binding affinities. Expet Opin Drug Discov 2015; 10(5): 449–461.

51.

Chaves

Sacramento

Ferreira

, et al. Atazanavir is a competitive inhibitor of SARS-CoV-2 Mpro, impairing variants replication in vitro and in vivo. Pharmaceuticals 2022; 15(1): 21.

52.

Abhinand

Nair

Krishnamurthy

, et al. Potential protease inhibitors and their combinations to block SARS-CoV-2. J Biomol Struct Dyn 2022; 40(2): 903–917.

53.

Fintelman-Rodrigues

Sacramento

Ribeiro Lima

, et al. Atazanavir, alone or in combination with ritonavir, inhibits SARS-CoV-2 replication and proinflammatory cytokine production. Antimicrob Agents Chemother 2020; 64(10): e00825. DOI: 10.1128/aac.00825-20.

54.

Metwaly

Elkaeed

Alsfouk

, et al. The computational preventive potential of the rare flavonoid, patuletin, isolated from Tagetes Patula, against SARS-CoV-2. Plants 2022; 11(14): 1886.

55.

Metwaly

Saleh

Alsfouk

, et al. In silico and in vitro evaluation of the anti-virulence potential of patuletin, a natural methoxy flavone, against Pseudomonas aeruginosa. PeerJ 2024; 12: e16826.

56.

Metwaly

Saleh

Alsfouk

, et al. Anti-virulence potential of patuletin, a natural flavone, against Staphylococcus aureus: in vitro and in silico investigations. Heliyon 2024; 10(2): e24075.

57.

Visualizer

. Discovery Studio Visualizer. San Diego, CA: Accelrys Software Inc, 2005, vol 2.

58.

Metwaly

El-Fakharany

Alsfouk

, et al. Comprehensive structural and functional analysis of patuletin as a potent inhibitor of SARS-CoV-2 targeting the RNA-dependent RNA polymerases. J Mol Struct 2024; 1311: 138424.

59.

Kim

Iyer

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

60.

Brooks

3rd Mackerell

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

61.

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 Theor Comput 2016; 12(1): 405–413.

62.

Best

Zhu

Shim

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

63.

Phillips

Braun

Wang

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

64.

Valdés-Tresanco

Valiente

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

65.

Bouysset

Fiorucci

. ProLIF: a library to encode molecular interactions as fingerprints. J Cheminf 2021; 13(1): 72.

66.

Salentin

Schreiber

Haupt

, et al. PLIP: fully automated protein-ligand interaction profiler. Nucleic Acids Res 2015; 43(W1): W443–W447.

67.

Maisuradze

Leitner

. Free energy landscape of a biomolecule in dihedral principal component space: sampling convergence and correspondence between structures and minima. Proteins 2007; 67(3): 569–578.

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

0.49 MB