Molecular docking simulation
Focusing on specific treatment, we chose the crystal structures of the wild type SARS-CoV-2 PLPro (PDB code: 7JRN),40 and SARS-CoV-2 Mpro (PDB code: 7CWC),41 respectively. For the other two enzymes, because the Spike protein might be for prevention and RdRP for the broad-spectrum treatment, which could also be helped by miR2911,32 the binding sites of the PLPro and Mpro were identified. As shown in Figure 3, the binding site in PLPpro was that of the original ligand GRL0617, while the binding sites in Mpro were complex. The basic site was that of ligand N3 together with another five binding sites. Sites I and II were separated from the N3-binding site, which was near the alpha (α)-helix-rich region of the Mpro; Sites III and IV were independent in the β-sheet-rich region, while Site V was at the linking position of the two regions. All of the natural products could have binding sites according to the steric and electronic surroundings. We chose the CDOCKER protocol considering the structural diversity of the ligands. Accordingly, the “-CDOCKER Interaction Energy” was selected as the basic index to evaluate the possibility of the interaction between each ligand and the binding site, while the “-CDOCKER Energy” was chosen as the reference index to evaluate the possible steadiness of the ligand-receptor complex.
From the simple object to the complex one, we chose to analyze the results from SARS-CoV-2 PLPro. Among the 482 natural products, 438 were docked into this site while 255 reached the basic requirement of potential interactions (we set as -CDOCKER Interaction Energy >30.0 kcal/mol). We listed the MOL IDs of the 255 compounds in Table S2 and illustrated the 2D binding patterns of the top 15 hits in Figure S1. The top 15 hits were also compared in Table 1. First, Shuanghuanglian seemed to have potential for SARS-CoV-2 PLPro because 90.87% (438 out of 482) of its investigated components were docked into the PLPro, which was higher than that of the compounds in the FDA-approved Drug Library supported by Selleck (China) (Shanghai, China) (81.14%; 2,229 out of 2,747). Simultaneously, the “-CDOCKER Interaction Energy” of the top hits were almost in the rational range of 40.0–60.0 kcal/mol, which was almost the most potential situation of CDOCKER before further modification and evaluation. Secondly, the flavones, glycosides, and polyphenols were preferred in the top hits, while long chain fatty acid esters also appeared. Specially, among the top hits, long chain fatty acid esters were all from Huang. Thirdly, Shuang and Lian were more important than Huang for binding to the PLPro. Additionally, the repeated compounds did not appear frequently (two out of 15; 13.33%). In particular, as shown in the 3D binding pattern of MOL003130 (Fig. 4a) and MOL002037 (Fig. 4b), the major key residues for the hydrogen bonds were Lys157, Leu162, Asp164, Arg166, Glu167, Gly266, Asn267, Tyr268, Gln269, Tyr273, and Thr301, respectively. We then chose the top three to evaluate their possible druggability.
Table 1The comparison of the top 15 hits in SARS-CoV-2 PLPro
MOL ID | -CDOCKER interaction energy (kcal/mol) | -CDOCKER energy (kcal/mol) | Source | Number of H-bonds | General name |
---|
MOL003130 | −6.29035 | 63.3313 | Shuang | 10 | Madreselvin A |
MOL002037 | 41.0151 | 60.9759 | Lian | 7 | Amentoflavone |
MOL000010 | −3.98465 | 58.1761 | Shuang | 7 | Rhoifolin |
MOL001875 | 44.3396 | 57.8812 | Shuang | 4 | Isochlorogenic acid |
MOL003051 | −1.68697 | 56.8953 | Shuang | 1 | Scolymoside |
MOL003309 | 38.3745 | 56.5246 | Lian | 6 | Plantainoside A |
MOL003076 | 35.8932 | 56.4097 | Shuang | 2 | 3,5-Di-O-caffeoylquinic acid methyl ester |
MOL003284 | 31.5003 | 56.2757 | Lian | 5 | Caleolarioside A |
MOL003077 | 38.4369 | 56.182 | Shuang | 8 | 4,5-Di-O-caffeoylquinic acid methyl ester |
MOL003334 | −3.93747 | 55.5806 | Lian | 7 | Forsythoside D |
MOL000415 | −2.95525 | 55.0801 | Shuang & Lian | 3 | Rutin |
MOL013161 | 51.9647 | 54.8761 | Huang | 1 | Methyl Hexacosanoate |
MOL007792 | 8.65799 | 54.3395 | Huang | 4 | Isomartynoside |
MOL009734 | 49.455 | 53.8073 | Huang | 1 | Methyl lignocerate |
MOL000007 | 14.4816 | 53.6873 | Shuang & Huang | 4 | Cosmetin |
Actually, with higher homology with SARS-CoV, Mpro has been widely investigated.26 Here, we also analyzed the results in the N3-binding site before that of the further defined ones. Among the 482 natural products, 451 were docked into this site, while 246 could be potential for interactions with the Mpro. We listed the MOL ID of the 246 compounds in Table S3 and illustrated the 2D binding patterns of the top 15 hits in Figure S2. The hints from the docking results in SARS-CoV-2 Mpro were different (Table 2). Initially, compared with that in PLPro, the success rate of docking was even as high as 93.57% (451 out of 482), which was also higher than that of the compounds in the FDA-approved Drug Library (83.25%; 2,287 out of 2,747). However, one questionable point was that the “-CDOCKER Interaction Energy” of the top hits was all beyond 60.0 kcal/mol, thus suggesting that the N3-binding site could be too large for the investigated ligands. Accordingly, we subsequently divided it into smaller sites. Secondly, flavones, glycosides, and polyphenols were preferred. Simultaneously, the long chain species were alkanes. The long chain compounds were all from Huang, and they were not favorable for generating hydrogen bonds. Thirdly, though Shuang and Lian still covered a majority of the top hits, their compounds were quite different from those docked in the PLPro because only two out of 15 (MOL003130 and MOL007792) were the same. All the top hits were unique herbs. We illustrated the 3D patterns of MOL003008 (Fig. 4c) and MOL003337 (Fig. 4d) to infer the key residues for the hydrogen bonds as Phe3, Arg4, Lys5, Arg131, Thr199, Ser284, Leu287, Glu288, and Glu290, respectively.
Table 2The comparison of the top 15 hits in SARS-CoV-2 Mpro at the N3-binding site
MOL ID | -CDOCKER interaction energy (kcal/mol) | -CDOCKER energy (kcal/mol) | Source | Number of H-bonds | General name |
---|
MOL003008 | 17.0653 | 74.3945 | Shuang | 9 | Madreselvin B |
MOL003337 | 13.6612 | 74.0027 | Lian | 6 | Forsythoside F |
MOL003332 | 29.1808 | 71.6769 | Lian | 7 | Forsythoside C |
MOL003110 | −52.6021 | 71.2342 | Shuang | 5 | Centauroside |
MOL003130 | 0.355922 | 68.1028 | Shuang | 7 | Madreselvin A |
MOL000870 | 54.4467 | 64.6227 | Huang | 0 | Hexatriacontane |
MOL007792 | 20.8246 | 63.9904 | Huang | 5 | Isomartynoside |
MOL003331 | 24.3554 | 63.6139 | Lian | 4 | Forsythiaside |
MOL003285 | −5.4399 | 63.3833 | Lian | 9 | N/A |
MOL003333 | 27.1839 | 62.5565 | Lian | 7 | Acteoside |
MOL003316 | 28.799 | 61.8015 | Lian | 5 | β-Hydroxyacteoside |
MOL003013 | 12.6776 | 61.1352 | Shuang | 4 | Secologanic dibutylacetal |
MOL003313 | 13.6329 | 60.7986 | Lian | 4 | Suspensaside A |
MOL000522 | 14.5554 | 60.2791 | Lian | 4 | Arctiin |
MOL005224 | 57.0222 | 60.1095 | Huang | 0 | Tetratetracintane |
In addition, it could be possible that smaller binding sites could be more rational than the N3-binding site for the prepared ligands. Instead of simply reducing the radius, we identified the new sites according to the real cavities. Preliminarily, we compared the feasibility of the five divided sites. The successful docked and interaction possible ligands were Site I: 339 and 21; Site II: 436 and 257; Site III: 390 and 173; Site IV: 367 and 169; Site V: 412 and 221. Because the top hit (MOL003283) in Site I indicated the “-CDOCKER Interaction Energy” of merely 36.0175 kcal/mol, we reasonably believed that Site I did not have the potential for the prepared ligands from Shuanghuanglian. The top 50 hits in each site were listed in Table S4, and the 2D binding patterns of the top 15 hits were illustrated in Figures S3-S6. Site II was similar too, but smaller than the N3-binding site. From the energy index in Table 3, Site II was also more rational than the N3-binding site, which realized our purpose of defining the divided sites. The top hits were all formed from flavones, glycosides, and polyphenols. Shuang and Lian contributed to 14 out of 15 top hits, except the one that appeared in both herbs. Since Site II was similar to the N3-binding site, a high percentage of ligands (33.3%, five out of 15, were marked in Table 3) appeared in the top hits in both sites. We compared the detailed binding patterns of MOL003130 in both sites (Fig. 5a, b, and d), and found that these two conformations interacted with the different residues in the Mpro. Coincidently, MOL003130 also appeared in the top hits in PLPro. Other repeated ligands in both the Mpro Site II and PLPro included MOL000415, MOL001875, MOL003051, MOL000010, and MOL003334. For each repeat, we picked the top three from the previously unpicked ones for further evaluation. They were MOL003316, MOL003313, MOL003013, MOL000415, MOL001875, and MOL003051.
Table 3The comparison of the top 15 hits in the SARS-CoV-2 Mpro Site II
MOL ID | -CDOCKER interaction energy (kcal/mol) | -CDOCKER energy (kcal/mol) | Source | Number of H-bonds | General name |
---|
MOL000415 | 6.83568 | 64.3218 | Shuang & Lian | 11 | Rutin |
MOL003130 | −8.23777 | 63.2077 | Shuang | 8 | Madreselvin A |
MOL001875 | 34.2741 | 61.9011 | Shuang | 7 | Isochlorogenic acid |
MOL003051 | 1.79757 | 61.1418 | Shuang | 7 | Scolymoside |
MOL002921 | 11.635 | 59.2781 | Huang | 9 | N/A |
MOL003316 | 21.7516 | 59.1990 | Lian | 7 | β-Hydroxyacteoside |
MOL000010 | 5.53976 | 59.0876 | Shuang | 6 | Rhoifolin |
MOL003118 | 37.2258 | 58.1035 | Shuang | 9 | Isochlorogenic acid C |
MOL003332 | 15.5971 | 58.0757 | Lian | 5 | Forsythoside C |
MOL003334 | 12.9183 | 56.2545 | Lian | 8 | Forsythoside D |
MOL006370 | 30.5446 | 55.3136 | Huang | 7 | 5-O-Caffeoylquinic acid |
MOL003313 | 4.35655 | 55.2171 | Lian | 4 | Suspensaside A |
MOL003013 | −17.4498 | 55.1203 | Shuang | 3 | Secologanic dibutylacetal |
MOL003010 | 22.1107 | 54.5034 | Shuang | 6 | Quercetin-3-O-β-D-glu |
MOL003336 | 15.8194 | 53.9310 | Lian | 8 | Forsythoside E |
When we divided the sites, we thought that Sites III and IV were not typical for the small molecular ligands because they were surrounded mainly by the β-sheet and their successfully docked ligands were relatively fewer (<400). Though the energy index was also less favorable, these two sites could have potential because the “-CDOCKER Interaction Energy” of the top hits was almost in the rational range of 40.0–60.0 kcal/mol (Tables 4 and 5). Interestingly, the compounds from Huang, regardless of their chain lengths, appeared more frequently in the top hits. Accordingly, we could not ignore the possibility that Huang might interact at these binding sites and act as an important supplement to the overall activity of Shuanghuanglian.
Table 4The comparison of the top 15 hits in the SARS-CoV-2 Mpro Site III
MOL ID | -CDOCKER interaction energy (kcal/mol) | -CDOCKER energy (kcal/mol) | Source | Number of H-bonds | General name |
---|
MOL000870 | 42.1147 | 48.0156 | Huang | 0 | Hexatriacontane |
MOL008595 | 42.8505 | 46.4620 | Huang | 1 | Methyl henicosanoate |
MOL005368 | 41.4543 | 45.0270 | Huang | 2 | Methyl tricosanoate |
MOL003284 | 27.3505 | 44.9463 | Lian | 3 | Caleolarioside A |
MOL003290 | 19.0479 | 44.4435 | Lian | 3 | N/A |
MOL003010 | 16.5869 | 43.5525 | Shuang | 6 | Quercetin-3-O-β-D-glu |
MOL000702 | 10.2468 | 43.2031 | Lian | 5 | Guajavarin |
MOL009730 | 29.8744 | 42.6731 | Huang | 1 | Methyl icos-11-enoate |
MOL000663 | 43.3447 | 42.6685 | Shuang | 1 | Lignoceric acid |
MOL002879 | 41.0219 | 42.5431 | Huang | 1 | Diop |
MOL003322 | −1.69405 | 42.5174 | Lian | 4 | Forsythinol |
MOL005224 | 31.3177 | 42.3766 | Huang | 0 | Tetratetracintane |
MOL002027 | 38.0839 | 42.3339 | Huang | 0 | Methyl behenate |
MOL003030 | 33.7532 | 42.3317 | Shuang | 0 | Ginnol |
MOL002934 | 16.9542 | 42.2364 | Huang | 5 | Neobaicalein |
Table 5The comparison of the top 15 hits in the SARS-CoV-2 Mpro Site IV
MOL ID | -CDOCKER interaction energy (kcal/mol) | -CDOCKER energy (kcal/mol) | Source | Number of H-bonds | General name |
---|
MOL003309 | 32.9477 | 55.5423 | Lian | 6 | Plantainoside A |
MOL000009 | 18.7637 | 51.3359 | Shuang | 5 | Luteolin-7-O-glucoside |
MOL003345 | 13.0808 | 50.3323 | Lian | 5 | N/A |
MOL000561 | 15.2567 | 50.2364 | Shuang & Lian | 4 | Astragalin |
MOL000437 | 15.4088 | 49.6075 | Lian | 5 | Hirsutrin |
MOL000007 | 15.1167 | 49.0453 | Shuang & Huang | 5 | Cosmetin |
MOL002935 | 15.7847 | 48.0702 | Huang | 4 | Baicalin |
MOL002912 | 22.8402 | 47.9820 | Huang | 3 | Dihydrobaicalin |
MOL002931 | 25.8930 | 47.8252 | Huang | 4 | Scutellarin |
MOL003284 | 30.3906 | 46.1796 | Lian | 4 | Caleolarioside A |
MOL003297 | −5.35172 | 46.0809 | Lian | 3 | N/A |
MOL003010 | 18.5803 | 45.6709 | Shuang | 6 | Quercetin-3-O-β-D-glu |
MOL000702 | 9.11693 | 44.8240 | Lian | 3 | Guajavarin |
MOL003128 | 2.8717 | 44.7907 | Shuang | 3 | Dinethylsecologanoside |
MOL003119 | −8.58632 | 44.5271 | Shuang | 6 | Loniceracetalide A |
Although the surrounding of Site V contained both the α-helix and β-sheet, the top hits seemed more similar to the ones in Site II, which were in an α-helix-rich region. As shown in Table 6, four out of the top five hits (in total five out of 15 as marked) repeated the top hits in Site II or the N3-binding site. We noticed that MOL003130 appeared again in the leading position with the potential energy values. Given that Site V was close to the N3-binding site, we illustrated the detailed binding patterns of MOL003130 in Site V (Fig. 5c) and compared its relative position in Mpro (Fig. 5d). Because these two conformations were both bound near the center of the corresponding sites, there seemed no key residues to interact with both of the conformations of MOL003130. Alternatively, MOL003130 could bind to the different sites of the Mpro simultaneously. We also selected the top three unpicked ligands from this site (MOL003336, MOL003284, and MOL003334).
Table 6The comparison of the top 15 hits in the SARS-CoV-2 Mpro Site V
MOL ID | -CDOCKER interaction energy (kcal/mol) | -CDOCKER energy (kcal/mol) | Source | Number of H-bonds | General name |
---|
MOL003130 | −3.17177 | 62.6809 | Shuang | 6 | Madreselvin A |
MOL003336 | −3.49283 | 57.1383 | Lian | 7 | Forsythoside E |
MOL003284 | 34.4530 | 56.6149 | Lian | 8 | Caleolarioside A |
MOL003334 | 2.75989 | 55.4900 | Lian | 8 | Forsythoside D |
MOL003285 | −13.2275 | 54.0015 | Lian | 4 | N/A |
MOL000536 | 13.0390 | 53.3301 | Lian | 3 | Matairesinoside |
MOL003309 | 31.5105 | 52.9643 | Lian | 6 | Plantainoside A |
MOL002702 | 49.0094 | 52.6923 | Shuang | 1 | Nonacosanol |
MOL003030 | 48.1793 | 52.3531 | Shuang | 2 | Ginnol |
MOL003020 | 7.01853 | 51.5526 | Shuang | 3 | Secologanoside 7-methylester |
MOL003118 | 31.1981 | 51.4023 | Shuang | 6 | Isochlorogenic acid C |
MOL009734 | 50.1233 | 51.1693 | Huang | 1 | Methyl lignocerate |
MOL003327 | 17.7518 | 51.0608 | Lian | 5 | Rengyoside C |
MOL000357 | −50.5619 | 50.5453 | All | 2 | Sitogluside |
MOL003292 | −11.9878 | 50.4217 | Lian | 3 | (+)-Epipinoresinol-4′-O-D-glucoside |