3.1. Optimized structures of the main protease–ligand complexes
In our previous molecular simulations [23], we considered 12 natural compounds extracted from Moringa oleifera and investigated their binding affinities against Mpro. Their chemical properties and structures are shown in Table S1 and Figure S1 of the Supplementary material (SM). In addition, in the present study, we considered 14 compounds from Aloe vera and 9 compounds from Nyctanthes arbor-tristis as candidate inhibitors against Mpro. Their pharmacokinetic properties evaluated using the SwissADME web tool [33] are listed in Tables S2 and S3 of SM and their chemical structures are shown in Figs. 1 and 2, respectively.
At first, the structures of the compounds were optimized, and the RESP charges were evaluated by G09 [34]. Using the optimized structure and RESP charge, each compound was docked into the binding site of the Mpro using the AutoDock4.2.6 program [40]. The candidate structures of the Mpro–ligand complex produced by the AutoDock4.2.6 program [40] were grouped into several clusters according to their structural similarities, and the clusters were ranked based on the binding energy between the Mpro and the compound. We selected three clusters with the largest number of poses because there was a higher probability that the compounds possessed as one of the candidate structures in these clusters. Therefore, we chose representative structures from these clusters and optimized them with explicit water molecules using the MM method of AMBER12 [41]. For the optimized structures, the total IFIEs between the compounds and each of the Mpro residues were precisely evaluated using the ab initio FMO method [45] to determine which compound bound most strongly to the Mpro.
The results of the docking, MM, and FMO calculations for the compounds from Aloe vera and Nyctanthes arbor-tristis are listed in Tables 1 and 2, respectively, where the lowest binding energy (BE) in kcal/mol, number of candidate poses, and Mpro residues involved in hydrogen bonds with each compound are included for the selected clusters. In the last column of these tables, the total IFIE in kcal/mol for each compound with all Mpro residues, evaluated using the ab initio FMO method is listed. Notably, Tables 1 and 2 indicate that the top-ranked cluster created by AutoDock4.2.6 did not necessarily correspond to the most stable structure in the FMO calculations. For example, as listed in Table 1, cluster 4 of the compound A1 had the highest total IFIE, indicating that A1 was bound most strongly to the Mpro in the conformation of the cluster 4, not in the cluster 1. Therefore, the tables suggested that the selection of clusters from the AutoDock4.2.6 results had to be done with caution. In the following analysis, for each compound, we used the structure with the highest total IFIE.
Table 1
This table shows the compounds extracted from Aloe vera with their lowest binding energy (BE) (kcal/mol), the number of poses, and the Mpro residues involved in H-bonds for the selected clusters obtained using the AutoDock4.2.6 program [40]. The 256 poses created were classified into several clusters based on the structural similarity, and each cluster was ranked in the order of the BE between the Mpro and the compound. We selected three clusters with the highest number of poses and evaluated the total inter-fragment interaction energy (IFIE) (kcal/mol) of each compound with all Mpro residues using the FMO method. These values are listed in the last column.
Compound
|
Cluster
|
BE
|
Poses
|
Residues involved in H-bonds
|
IFIE
|
A1
|
1
|
−3.31
|
32
|
No H-bond
|
−22.5
|
2
|
−3.19
|
75
|
Gln189
|
−43.4
|
4
|
−2.89
|
108
|
Cys145
|
−57.0
|
A2
|
1
|
−3.58
|
78
|
Ser144
|
−53.8
|
2
|
−3.34
|
38
|
Leu141, Cys145, Glu166
|
−101.1
|
3
|
−3.26
|
87
|
Asn142, Met165, Glu166
|
−96.2
|
A3
|
1
|
−3.43
|
65
|
Glu166
|
−97.0
|
2
|
−3.04
|
70
|
Cys145, Glu166, Leu167
|
−99.4
|
13
|
−2.4
|
35
|
Asn142
|
−36.0
|
A4
|
1
|
−3.52
|
243
|
Asn142, Glu166
|
−70.0
|
2
|
−3.29
|
1
|
Hie164, Met165
|
−41.1
|
3
|
−3.20
|
12
|
Thr24, Thr25, Thr45, Ser46
|
−54.6
|
A5
|
1
|
−4.00
|
174
|
Glu166
|
−55.2
|
3
|
−3.72
|
62
|
Thr190, Ala191
|
−39.9
|
4
|
−3.65
|
11
|
Gln192, Ala193
|
−49.2
|
A6
|
1
|
−3.61
|
148
|
Asn142, Glu166
|
−106.1
|
2
|
−3.46
|
93
|
Asn142, Ser144, Glu166
|
−116.8
|
3
|
−3.23
|
15
|
Glu166, Leu167, Arg188, Gln189
|
−70.8
|
A7
|
1
|
−6.02
|
234
|
No H-bond
|
−41.4
|
2
|
−5.78
|
2
|
No H-bond
|
−40.5
|
3
|
−5.67
|
20
|
Met17, Val18
|
−59.1
|
A8
|
1
|
−5.10
|
200
|
No H-bond
|
−51.0
|
3
|
−4.86
|
22
|
Hie164, Met165, Glu166
|
−94.5
|
4
|
−4.70
|
28
|
Hid41
|
−51.4
|
A9
|
1
|
−4.80
|
161
|
Leu141, Asn142, Glu166
|
−78.1
|
4
|
−4.31
|
40
|
Cys145, Glu166
|
−56.5
|
6
|
−4.22
|
36
|
No H-bond
|
−50.6
|
A10
|
1
|
−4.41
|
229
|
Cys145, Glu166, Hid172
|
−112.1
|
4
|
−3.92
|
20
|
Thr26, Leu27
|
−78.6
|
5
|
−3.87
|
3
|
Asn142, Glu166
|
−90.6
|
A11
|
1
|
−4.67
|
115
|
Asn142, Ser144, Glu166, His172
|
−95.7
|
3
|
−4.33
|
55
|
No H-bond
|
−55.9
|
5
|
−4.03
|
32
|
Hid41, Asn142, Gln189
|
−71.7
|
A12
|
1
|
−5.05
|
167
|
Hid41, Asn142, Gln189
|
−91.2
|
2
|
−4.93
|
27
|
Asn142, Glu166, Gln189
|
−122.6
|
3
|
−4.81
|
39
|
Glu166, Gln189
|
−78.8
|
A13
|
1
|
−5.26
|
136
|
Glu166
|
−70.0
|
3
|
−5.20
|
32
|
Thr26
|
−69.9
|
5
|
−4.75
|
31
|
Cys145
|
−73.8
|
A14
|
1
|
−4.75
|
39
|
Glu166, Gln189
|
−111.5
|
6
|
−4.30
|
23
|
Thr26, Leu27, Asn142, Glu166, Leu167
|
−150.1
|
12
|
−4.07
|
24
|
Thr24, Thr25, Asn142
|
−102.9
|
Table 2
This table shows the compounds extracted from Nyctanthes arbor-tristis with their lowest binding energy (BE) (kcal/mol), the number of poses, and the Mpro residues involved in H-bonds for the selected clusters obtained using the AutoDock4.2.6 program [40]. The 256 poses created were classified into several clusters based on the structural similarity, and each cluster was ranked in the order of the BE between the Mpro and the compound. We selected three clusters with the highest number of poses and evaluated the total IFIE (kcal/mol) of each compound with all the Mpro residues using the FMO method. These values are listed in the last column.
Compound
|
Cluster
|
BE
|
Poses
|
Residues involved in H-bonds
|
IFIE
|
N1
|
1
|
−3.75
|
85
|
No H-bond
|
−43.2
|
5
|
−3.24
|
17
|
No H-bond
|
−43.6
|
9
|
−3.04
|
32
|
Glu166
|
−36.7
|
N2
|
2
|
−2.98
|
100
|
Glu166
|
−75.5
|
3
|
−2.96
|
44
|
No H-bond
|
−24.3
|
5
|
−2.87
|
38
|
No H-bond
|
−33.4
|
N3
|
1
|
−3.69
|
53
|
Glu166
|
−82.8
|
2
|
−3.66
|
47
|
Leu141, Glu166
|
−87.0
|
4
|
−3.45
|
47
|
Thr26
|
−57.7
|
N4
|
1
|
−4.54
|
224
|
No H-bond
|
−28.7
|
2
|
−4.54
|
3
|
No H-bond
|
−23.3
|
3
|
−4.48
|
29
|
No H-bond
|
−33.6
|
N5
|
1
|
−2.87
|
78
|
No H-bond
|
−25.3
|
2
|
−2.78
|
160
|
Glu166
|
−56.0
|
3
|
−2.45
|
23
|
Asn142
|
−40.2
|
N6
|
1
|
−3.31
|
67
|
Gln189
|
−41.6
|
2
|
−3.13
|
46
|
Glu166
|
−57.2
|
4
|
−3.05
|
86
|
Gln189
|
−32.8
|
N7
|
1
|
−4.9
|
78
|
No H-bond
|
−50.0
|
2
|
−4.69
|
74
|
Asn142, Gly143
|
−64.7
|
3
|
−4.66
|
47
|
Asn142, Gly143
|
−66.9
|
N8
|
1
|
−6.35
|
255
|
No H-bond
|
−173.7
|
2
|
−5.49
|
1
|
No H-bond
|
−109.6
|
N9
|
1
|
−4.85
|
196
|
His172
|
−74.5
|
2
|
−4.65
|
19
|
Gly143, Glu166
|
−94.8
|
3
|
−4.40
|
17
|
Thr26, His172
|
−75.0
|
3.2. Binding affinities of the compounds and the main protease
The total IFIEs of the Mpro residues and the compounds extracted from Moringa oleifera [23], Aloe vera, and Nyctanthes arbor-tristis were compared, as shown in Fig. 3. The red bars indicate the compounds with the highest total IFIE from each plant. It should be noted that anthocyanin (N8) was excluded from the list of Mpro inhibitors because it had a positive charge and interacted differently with the Mpro residues compared with that of the other compounds.
As described in our previous study [23], of the 12 compounds from Moringa oleifera, niaziminin (M9) was found to bind more strongly to the Mpro (Table S4 of SM). This compound interacted strongly with the negatively charged Glu166 residue, resulting a strong binding to the Mpro. The total IFIE was evaluated, having a magnitude of 136.5 kcal/mol, at least 17.5 kcal/mol higher than that of the other compounds extracted from Moringa oleifera. Therefore, we proposed novel compounds based on niaziminin (M9) to investigate whether the introduction of a hydroxyl group into niaziminin (M9) would enhance its binding affinity to the Mpro. Niaziminin (M9) is a thiocarbamate isolated from the leaves of Moringa oleifera [51]. The presence of an acetoxy group at the 4′-position of niaziminin (M9) is thought to be important for its ability to inhibit the activation of Epstein–Barr virus [52].
As indicated in Fig. 3, of the 34 remaining compounds after the exclusion of N8, compound feralolide (A14) from Aloe vera, had the highest total IFIE (150.1 kcal/mol). This result was approximately 13.6 kcal/mol greater than that of the compound with the total IFIE of the second-highest magnitude, i.e., niaziminin (M9) (136.5 kcal/mol) from Moringa oleifera. Aloesaponarin-I (A12) (122.6 kcal/mol) from Aloe vera had the total IFIE of the third-highest magnitude. Recently, Mpiana et al. [53] performed a molecular docking study with the Mpro using Gasteiger charges for 10 compounds extracted from Aloe vera, and the compound with the highest score in their study was feralolide (A14). Therefore, the results obtained by ab initio FMO calculations and presented here are comparable to their results, suggesting that feralolide (A14) could be a potent inhibitor of the Mpro and therefore a drug lead compound against COVID-19.
To elucidate the strong affinity between feralolide (A14) and the Mpro, we investigated the IFIEs of the compounds feralolide (A14), eupatorine (A13), and aloesaponarin-I (A12), as these compounds had similar structures, as shown in Fig. 1. Feralolide (A14) (Fig. 4a) bound most strongly to Glu166 and also interacted with other Mpro residues Leu27, Asn142, and Leu167, resulting in a total IFIE of high magnitude. Therefore, these residues were found to be the main contributors to the strong interaction between the Mpro and feralolide (A14). In contrast, aloesaponarin-I (A12) (Fig. 4c) interacted strongly with only Glu166 and weakly with Asn142 and Gln189, resulting in a total IFIE of lower magnitude with the Mpro. As shown in Fig. 4b, the IFIEs of eupatorine (A13) with the Mpro were noticeably different from those of feralolide (A14) and aloesaponarin-I (A12). There was no strong interaction between eupatorine (A13) and the Mpro residues, and only Cys145 interacted with eupatorine (A13) at a higher magnitude than 10 kcal/mol. Therefore, the total IFIE of eupatorine (A13) was of a significantly lower magnitude than that of feralolide (A14) and aloesaponarin-I (A12). The comparison of Figs. 4a, 4b, and 4c revealed that Glu166, Leu167, Leu27, and Asn142 were important for the strong binding of feralolide (A14) to the Mpro.
Therefore, we investigated the nature of the interactions between Mpro residues and feralolide (A14), eupatorine (A13), aloesaponarin-I (A12) to understand which functional groups were responsible for the binding of Mpro. As shown in Fig. 5a, two hydroxyl groups in the upper end of feralolide (A14) formed strong hydrogen bonds (1.51 and 1.59 Å) with Glu166 and the peptide backbone between Glu166 and Leu167. In addition, the carbonyl group of feralolide (A14) formed a hydrogen bond with the amide group of Asn142, while the hydroxyl group formed a hydrogen bond with the peptide backbone between Thr26 and Leu27. Aloesaponarin-I (A12) (Fig. 5c) formed only one strong hydrogen bond (1.57 Å) with Glu166. In contrast, as shown in Fig. 5b, there was no hydrogen bond interaction between eupatorine (A13) and the Mpro residues, although there were only weak electrostatic interactions with Thr26, Gly143, and Cys145. Conversely, Fig. 5a indicates that feralolide (A14) possessed many groups for strong binding with the Mpro residues. In particular, the two hydroxyl groups in the upper end effectively strengthened the interaction between feralolide (A14) and the Mpro residues. Therefore, feralolide (A14) was expected to be a drug lead compound for the development of potent inhibitors against COVID-19.
Aloesaponarin-I (A12) and aloesaponarin-II (A8), shown in Fig. 1, are anthraquinones isolated from the roots of Aloe vera, and their antiviral activity against the human influenza virus was demonstrated in a previous study [54]. In addition, Parvez et al. [55] tested the anti-hepatitis B virus (HBV) potential of Aloe vera extract and its anthraquinones in hepatoma cells. They reported that the effect of aloe-emodin (A10) is comparable with that of lamivudine (a nucleoside reverse transcriptase inhibitor approved for the treatment of HBV) and appears to be the most promising natural anti-HBV drug, with CYP3A4 activation enhancing its therapeutic efficacy. Our present results (Fig. 3) indicated that aloesaponarin-I (A12) had the second-highest total IFIE of the compounds from Aloe vera, while aloesaponarin-II (A8) had an average-sized total IFIE. Therefore, it was expected from the total IFIE evaluated by the FMO method that aloesaponarin-I (A12) could also be a potent inhibitor of the Mpro.
In addition, we analyzed the interactions of the other compounds with a total IFIE > 100 kcal/mol with the Mpro. As shown in Fig. 3, the Aloe vera compounds caffeic acid (A2), esculetin (A6), aloe-emodin (A10), aloesaponarin-I (A12), and feralolide (A14) came into this category. Interestingly, all these compounds formed a hydrogen bond with Glu166, and the magnitude of the IFIE of each compound with Glu166 was > 50 kcal/mol. Since esculetin (A6), a coumarin derivative found in various medicinal plants, inhibits proliferation and induces apoptosis in many types of human cancer cells, it is a promising chemotherapeutic agent [56]. In addition, esculetin (A6) has been widely used as an anti-inflammatory, antioxidant, antibacterial, and anti-diabetic therapeutic [57]. Recently, the anti-HBV activity of esculetin (A6) was investigated, and it was shown to inhibit HBV replication effectively both in vitro and in vivo [58]. Here, of the 14 compounds from Aloe vera, esculetin (A6) exhibited the IFIE of the highest magnitude (73 kcal/mol) with Glu166 and formed two hydrogen bonds. These results indicated that hydrogen bonding to Glu166 is a key for Aloe vera compounds to bind strongly to the Mpro. According to Juárez-Saldívar et al.[59], Glu166 is important for maintaining the shape of the Mpro active site. In addition to this hydrogen bond, most of the Aloe vera compounds participate in electrostatic interactions with Gly143, Ser144, and Cys145. In contrast, Nguyen et al. [60] found that Gly143 of Mpro is the most important residue for hydrogen bonding with ligands, followed by Glu166, Cys145, and His163 [5].
Table 2 lists the results of the docking and FMO calculations for the nine compounds extracted from Nyctanthes arbor-tristis. From the two or three clusters for each compound, we selected the cluster with the total IFIE of the highest magnitude and compared the magnitude of this IFIE for each compound. As shown in Fig. 3, the magnitudes of the total IFIEs of the Nyctanthes arbor-tristis compounds, excluding N8, were significantly lower than those of the Aloe vera compounds. Actually, N8 had the highest total IFIE of all the compounds investigated here. However, this strong interaction came from the electrostatic interaction of the positive charge on N8 with the negatively charged residues of the Mpro, as shown in Fig. 6. Therefore, the electrostatic nature of this interaction of N8 with the Mpro was significantly different from that of the other compounds; thus, for the present study, we eliminated N8 from the list of candidate Mpro inhibitors.
Of the eight remaining compounds from Nyctanthes arbor-tristis, apigenin (N9) exhibited a total IFIE of high magnitude (94.8 kcal/mol). As indicated in Fig. 7a, apigenin (N9) interacted strongly only with Glu166 via electrostatic interactions, suggesting that Glu166 is a key residue for stabilizing bound apigenin (N9). However, no residue other than Glu166 interacted strongly with apigenin (N9), resulting in the lower magnitude of its total IFIE compared with that of feralolide (A14) and aloesaponarin-I (A12). Previously, the antiviral activity of the naturally occurring flavonoid apigenin (N9) was investigated in vitro and in vivo against several viruses including enterovirus 71, hepatitis C virus, human immunodeficiency virus, and adenoviruses [61]. Therefore, it was expected that apigenin (N9) could also be an inhibitor of the Mpro.
Long-chain fatty acids such as pelargonic acid (N2) and myristic acid (N3) were also compared. As indicated in Fig. 7b, myristic acid (N3) interacted strongly only with Glu166 and Leu141 of the Mpro and had a total IFIE of lower magnitude compared with that of apigenin (N9). Therefore, we considered these long-chain fatty acids unsuitable for the potent inhibitors of Mpro.
To explain the weaker binding of apigenin (N9) to the Mpro compared with that of feralolide (A14), we analyzed the binding interactions of these compounds with the Mpro residues in the binding site. As shown in Fig. 8b, three hydroxyl groups of feralolide (A14) formed strong hydrogen bonds with the Mpro residues, resulting in the total IFIE of the highest magnitude. In contrast, Fig. 8a indicates that only one hydroxyl group of apigenin (N9) formed a hydrogen bond with Glu166 although apigenin (N9) has a similar structure of feralolide (A14) and aloesaponarin-I (A12). As a result, the magnitude of the total IFIE of apigenin (N9) was significantly lower than that of feralolide (A14) and aloesaponarin-I (A12), as shown in Fig. 3. These results demonstrated that the number and position of hydroxyl groups affect the compound’s ability to interact with Mpro residues in the binding site.
3.3. Specific interactions of the compounds with the main protease
To elucidate the difference in the specific interactions of the 35 natural compounds contained in Moringa oleifera, Aloe vera, and Nyctanthes arbor-tristis with the Mpro residues, we classified these compounds based on their IFIEs using the visualized cluster analysis of protein-ligand interactions [62]. In this analysis, the ligands were classified into different groups depending on the similarity of the IFIEs of each ligand with the residues in the ligand-binding site. We eliminated N8 as a candidate inhibitor of the Mpro because it is positively charged and its way of interaction with the Mpro residues was significantly different from that of the other compounds.
As shown in Fig. 9, the compounds were classified into two main groups depending on the nature of their interaction with Glu166 in the Mpro binding site. Feralolide (A14), which had the total IFIE of the highest magnitude for all the compounds investigated here, was classified in the upper group shown in Fig. 9. Within this group, all compounds interacted strongly with Glu166, while the other compounds were completely separated from the feralolide (A14) subgroup. No other compounds had similar interactions with the Mpro residues compared to feralolide (A14). Among the 34 compounds analyzed here, feralolide (A14) had an IFIE of particularly high magnitude with the Mpro. In contrast, niaziminin (M9), which had the total IFIE of the second-highest magnitude as shown in Fig. 3, belonged to the same subgroup of aloesaponarin-I (A12). Therefore, Fig. 9 reveals that the IFIEs of niaziminin (M9) and aloesaponarin-I (A12) with the Mpro were similar although these compounds had different chemical structures, and both had the total IFIEs of significantly lower magnitude than that of feralolide (A14).
To understand the reason for these difference in IFIEs, we compared the interactions of these compounds with the key residues in the Mpro binding site. As shown in Fig. 5a, the hydroxyl groups at one end of feralolide (A14) formed hydrogen bonds with the negatively charged Glu166, whereas that at the other end formed hydrogen bond with the peptide backbone between Thr26 and Leu27. These hydrogen bonds of feralolide (A14) with two separate sites in the Mpro binding pocket were achieved by its vertical conformation in the pocket, as shown in Fig. 10a. Notably, of the compounds in this study, only feralolide (A14) was bound in this conformation. Feralolide (A14) also interacted electrostatically with the Mpro residues such as Gln189 and Asn142 in the binding site. In contrast, Fig. 10b indicates that relative to feralolide (A14), niaziminin (M9) and aloesaponarin-I (A12) were bound in a perpendicular conformation in the Mpro-inhibitor binding site, interacting with Glu166, Gln189, and Asn142. As these compounds did not interact with Thr26 and Leu27, their binding properties were not similar to those of feralolide (A14). The magnitude of the total IFIE of aloesaponarin-I (A12) with Mpro was the third-highest of the compounds in this study. Therefore, aloesaponarin-I (A12) is also expected to be a potent inhibitor against Mpro in addition to feralolide (A14) and niaziminin (M9).
3.4. Proposal of novel compounds based on feralolide (A14) and niazirinin (M12)
In our previous molecular simulations [23], we investigated the binding properties of 12 compounds extracted from Moringa oleifera with the Mpro and found that, niaziminin (M9) is bound most strongly to the Mpro. The chemical structures of the 12 compounds and the results of the simulations are shown in Figure S1 and Tables S1 and S2 of SM. Therefore, we proposed novel compounds based on niaziminin (M9) and revealed that the introduction of a hydroxyl group into niaziminin (M9) enhanced its binding to the Mpro.
In the present study, we first attempted to propose candidate potent inhibitors of the Mpro based on feralolide (A14), which had the total IFIE of the highest magnitude of the compounds studied here, as shown in Fig. 3. We replaced each hydrogen atom of feralolide (A14) with a hydroxyl group to propose nine novel compounds and investigated their binding properties with the Mpro in the same conformation of feralolide (A14). Their structures and total IFIEs with Mpro were evaluated using the FMO method. As shown in Table 3, the total IFIE was changed significantly depending on the hydroxylation site. In particular, the magnitude of the total IFIEs of feralolide derivatives (A14b), (A14d), and (A14i) were higher than 160 kcal/mol and at least 10 kcal/mol higher than that of feralolide (A14), indicating that the introduction of a hydroxyl group could significantly enhance the binding of feralolide derivatives to the Mpro.
Table 3. This table shows the total IFIE (kcal/mol) of feralolide (A14) and its proposed compounds with the Mpro, evaluated using the FMO method. The proposed compounds are defined as the compounds A14a−A14i based on the hydroxylation site. For example, in the compound (A14a), the hydrogen atom in the a-site of A14, as shown below, is replaced by a hydroxyl group.
To increase our understanding of this enhancement, we compared the IFIEs of each compound with the Mpro residues. As shown in Fig. 11a, feralolide (A14) interacted strongly with Glu166, Leu167, Leu27, and Asn142. By introducing a hydroxyl group in the d-site of feralolide (A14d), the IFIE between Glu166 and feralolide (A14) was further enhanced by 30 kcal/mol as shown in Fig. 11c, resulting in the total IFIE of the highest magnitude of 171.1 kcal/mol. In contrast, as shown in Figs. 11b and 11d, the introduction of a hydroxyl group in the b or i-site of feralolide (A14) enhanced the IFIE with Gln189 or Cys145 although the effect was not as strong as that seen with feralolide derivative (A14d). Therefore, hydroxylation in the d-site was found to be most effective for enhancing the binding between feralolide (A14) and the Mpro.
To understand this enhancement, we investigated the interaction of the proposed feralolide derivatives with the selected Mpro key residues. As shown in Fig. 12a, feralolide (A14) formed hydrogen bonds with Glu166, Asn142, and the peptide backbones between Glu166 and Leu167 and between Thr26 and Leu27. In the Mpro–feralolide derivative (A14d) complex (Fig. 12c), the extra hydroxyl group created an additional hydrogen bond with Glu166. As a result, the IFIE of Glu166 was significantly enhanced as shown in Fig. 11c. In contrast, Fig. 12b indicates that if the b-site was hydroxylated to produce feralolide derivative (A14b), this formed a new hydrogen bond with Gln189, leading to the enhancement of IFIE with Gln189. As the hydroxyl group of feralolide derivative (A14i) interacted electrostatically with Cys145, as shown in Fig. 12d, the IFIE of Cys145 was enhanced in feralolide derivative (A14i), as indicated in Fig. 11d. As mentioned above, depending on the site of hydroxylation on feralolide (A14), the interactions of the feralolide (A14) derivatives and specific residues of the Mpro were significantly different. Of the proposed feralolide (A14) derivatives, derivative (A14d) was found to have the total IFIE of the highest magnitude with the Mpro residues, indicating that it could be a candidate as a potent inhibitor of the Mpro.
Next, we proposed novel derivatives of niazirinin (M12) and investigated their binding properties to the Mpro because niazirinin (M12) had the total IFIE of the second-highest magnitude of the compounds extracted from Moringa oleifera, as shown in Fig. 3. One hydrogen atom of niazirinin (M12) was replaced by a hydroxyl group to create candidate compounds, and their total IFIEs with the Mpro were investigated. As indicated in Table S5 of SM, only niazirinin derivatives (M12c) and (M12d) had higher total IFIEs compared with niazirinin (M12). However, the magnitude of the total IFIEs of the niazirinin derivatives (M12c) and (M12d) was lower than that (150 kcal/mol) of feralolide (A14). Therefore, it was revealed that the derivatization of niazirinin (M12) by the introduction of a hydroxyl group was not as effective for enhancing the binding affinity to the Mpro, compared with the derivation of feralolide (A14).
To understand the change in the total IFIE is due to the introduction of an extra hydroxyl group into niazirinin (M12), we compared the IFIEs of niazirinin (M12) and its derivatives (M12c), (M12d) with the Mpro. As shown in Figure S2a of SM, niazirinin (M12) interacted strongly with Glu166 and Asn142. The effect of the hydroxylation in the c-site of M12 was, at most, 5 kcal/mol for Cys145 (Figure S2b). In contrast, the hydroxylation in the d-site significantly enhanced the interactions of niazirinin (M12) with the residues Leu141 and Glu166 (Figure S2c). However, the interactions of niazirinin (M12) with some residues around Leu141 were weakened by the hydroxylation, resulting only a small enhancement of the total IFIE for the niazirinin derivative (M12d).
Therefore, we investigated the nature of the interactions of Mpro key residues with niazirinin (M12) and derivatives (M12c) and (M12d). As shown in Figure S3a of SM, niazirinin (M12) formed a strong hydrogen bond with Glu166 and interacted electrostatically with Asn142 and Gly143. As shown in Figure S3b, the hydroxyl group introduced in the c-site of niazirinin (M12) interacted electrostatically with Cys145. In addition, a hydrogen bond was formed with a distance of 1.65 Å between the Glu166 side chain and a hydroxyl group of another site on the niazirinin derivative (M12c). The electrostatic interactions with the Leu141–Asn142–Gly143 peptide were also enhanced, as shown in Figure S3b. These changes in interactions in the binding site were considered to cause a significant (about 20 kcal/mol) increase in the total IFIE of the niazirinin derivative (M12c). In contrast, the hydroxyl group introduced in the d-site of niazirinin (M12d) formed a strong hydrogen bond with the peptide backbone between Phe140 and Leu141 (Figure S3c). In addition, an extra hydrogen bond was formed between the niazirinin derivative (M12d) and Glu166, resulting in the enhancement of the IFIE for Glu166. However, as shown in Figures S2b and S2c, the effect of introducing a hydroxyl group to niazirinin (M12) on the interactions between M12 and the Mpro was found to be at most 12 kcal/mol, which was less significant than that (at most 30 kcal/mol) for the hydroxylated feralolide (A14) derivatives shown in Fig. 11.
To elucidate which compound proposed in our study was bound more strongly to the Mpro, we compared their total IFIEs in Fig. 13. Niaziminin derivative (M9d) and feralolide derivatives (A14d), (A14b) had total IFIEs with a rather higher magnitude compared to those of the other compounds, indicating that they were expected to be the potent inhibitors of Mpro. To confirm the stability of binding of these compounds to the Mpro, we conducted 100-ns MD simulations for the complexes of Mpro with each of these compounds at 310 K, within an explicit water box. The results revealed that the compounds kept their binding conformations in the ligand-binding pocket of Mpro. Therefore, our proposed compounds are expected to be potent inhibitors of Mpro. The details of the MD results will be shown elsewhere.