Structural identification and properties of the RV capsid proteinThe obtained 3D structure of the RV capsid protein had four chains (chain1, chain2, chain3, and chain4), where chain1 was already determined as the VP1 protein; therefore, the other three chains were deleted using PyMOL, and only chain1 was selected for the molecular docking process (Fig. 1). The primary sequence of the target protein, the RV capsid protein, was obtained in FASTA format from the UniProt database under the accession number Q82122, where the complete genome codes for 2153 amino acid residues and the VP1 gene codes for 292 amino acids’ long proteins were stored. The physiochemical characterization of VP1 proteins using the ExPASy ProtParam tool determined that it contains a total of 30 negatively charged residues and 26 positively charged residues. Detailed information is given in Table S1. The functional analysis of target protein sequences using Interpro was used to determine the functional and conserved domains. However, protein family membership was not predicted (Fig. 2). Binding pocket(s) identification on the target protein was determined based on active sites present on the protein using CASTp. A total of 29 binding pockets were determined with different areas and volumes (Table 1).Figure 1Three-dimensional structure of rhinovirus capsid protein: a chain1 (VP1) is green, chain2 (VP2) is blue, chain3 (VP3) is pink, and chain4 (V4) is yellow, b. Chain1 (VP1) after deleting other chains.Figure 2Determination of functional and conserved domains in VP1 protein.Table 1 Description of binding pockets obtained in VP1 protein.Ligand selection, preparation, and analysis of their propertiesFirstly, a total of 30 ligands were selected for their antiviral properties against the target protein. These ligands were selected based on our own GC-MS data of S. rosmarinus essential oils from different regions of India, and the phenylpropanoid compounds of S. rosmarinus confirmed the presence of phenylpropanoid biosynthesis-specific genes through a KEGG pathway analysis of own transcriptome data. The selected ligands were β pinene, Camphor, Caryophyllene, Citral, Eocalyptol, Estragole, Isoborneol, Linalool, Neral, n hexadecanoic acid, α terpineol, γ terpinene, (4 methoxy benzyl) phenethyl amine, (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester, 2,6,10 dodecatrien 1 ol, (3,7,11 trimethyl), Benzenesulfonamide 4 methyl N (5 nitro 2 pyridinyl), Bornyl acetate, Carnosic acid, Citronellol, Diethyl phthalate, D limonene, (E,E,Z 1,3,12) nonadecatriene (5,14 diol), Geranic acid, Hydratropic acid undec 2 en 1 yl ester, Methyl abietate, Methyl eugenol, Rosefuran oxide, Rosmanol, Rosmarinic acid, and Shogaol. Antiviral drugs Pleconaril and Nitazoxanide were also selected as ligands for comparative studies. The 3D and 2D structures (See Figs. S1–S33) of all the selected ligands were downloaded in .sdf and .png file formats, respectively, from the PubChem database.The PkCSM tool was used as a first filter to identify whether the selected ligands followed Lipinski’s rule of five because this is the first criterion to check before considering any compound as a drug candidate. Only one rule was violated by (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester, n hexadecanoic acid, Hydratropic acid undec 2 en 1 yl ester, and Methyl abietate, each (Table 2). But, as per the rule, the drug likeliness of any compound or molecule is valid until it violates more than one criterion among the following:
A maximum of five hydrogen bond donors are allowed;
Maximum ten hydrogen bond acceptors;
A molecule with a mass of under 500 daltons;
An estimated octanol–water partition coefficient (ClogP) of not more than 5.
Table 2 Properties of selected rosemary compounds under Lipinski’s rule of five.Therefore, all selected ligands passed the test.Further, the selected ligands were screened by calculating the ADMET properties, which gives an idea about the pharmacokinetic properties of ligands using the same PkCSM tool. Detailed information about the ADMET properties of the selected ligands is given in Tables S2–S6.Molecular docking and analysis of protein–ligand complexThe Auto-Dock-vina-based online docking platform CB-Dock was used for blind docking between chain 1 (VP1) of the RV coat protein and the selected ligands of the rosemary. Simultaneously, the docking was also performed between the VP1 protein and selected antiviral drugs. The 3D structures of the target protein were uploaded and docked with the 3D structure of each ligand by uploading them individually. The CB-Dock results (Fig. 3) were based on the possible pose forms between receptor and ligands, and the best poses were selected based on the vina score, size of the cavity, grid map, etc. Here, the interaction models were developed based on protein–ligand binding affinity prediction using a curvature-dependent surface-area model47 between the VP1 protein and the different ligands. Binding energy values (vina score) were calculated, and the interaction with the lowest binding energy was chosen (Table 3). The interaction in the largest cavity was chosen during the docking when the binding energy was found to be the same. The least vina score predicted against the VP1 protein was shown by the Benzenesulfonamide 4 methyl N (5 nitro 2 pyridinyl). The top eight predicted ligands based on vina score cut off <−8 were Benzenesulfonamide 4 methyl N (5 nitro 2 pyridinyl) (−9.5 kcal/mol), Rosmarinic acid (−9.0 kcal/mol), (4 methoxy benzyl) phenethyl amine (−8.9 kcal/mol), Shogaol (−8.6 kcal/mol), Methyl abietate (−8.6 kcal/mol), Hydratropic acid undec 2 en 1 yl ester (−8.6 kcal/mol), (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester (8.1 kcal/mol), and 2,6,10 dodecatrien 1 ol, (3,7,11 trimethyl) (−8.0 kcal/mol).Figure 3CB-dock-based molecular docking analysis and interaction of VP1 protein with selected rosemary compounds: (i). β pinene, (ii). Camphor, (iii). Caryophyllene, (iv). Citral, (v). Eocalyptol, (vi). Estragole, (vii). Isoborneol, (viii). Linalool, (ix). Neral, (x). n hexadecanoic acid, (xi). α terpineol, (xii). γ terpinene, (xiii). (4 methoxy benzyl) phenethyl amine, (xiv). (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester, (xv). 2,6,10 dodecatrien 1 ol, (3,7,11 trimethyl), (xvi). Benzenesulfonamide 4 methyl N (5 nitro 2 pyridinyl), (xvii). Bornyl acetate, (xviii). Carnosic acid, (xix). Citronellol, (xx). Diethyl phthalate, (xxi). D Limonene, (xxii). (E,E,Z 1,3,12) Nonadecatriene (5,14 diol), (xxiii). Geranic acid, (xxiv). Hydratropic acid undec 2 en 1 yl ester, (xxv). Methyl abietate, (xxvi). Methyl eugenol, (xxvii). Rosefuran oxide, (xxviii). Rosmanol, (xxix). Rosmarinic acid, and (xxx). Shogaol.Table 3 Molecular docking results of selected rosemary compounds with VP1 protein.PyMol was used to visualize the interaction between protein and ligand complexes, whereas, the LigPlot+ was used to visualize and interpret bonds between protein–ligand complexes. The LigPlot+ generated 2D diagrams (Fig. 4) showing how ligands interacted with amino acids at the protein’s active site from the 3D coordinate of the protein–ligand complex saved in the .pdb file formats. The ligands that interacted via hydrophobic interaction only, and not a hydrogen bond, were β pinene, Camphor, Caryophyllene, Eocalyptol, n hexadecanoic acid, γ terpinene, (4 methoxy benzyl) phenethyl amine, 2,6,10 dodecatrien 1 ol, (3,7,11 trimethyl), Carnosic acid, D limonene, Hydratropic acid undec 2 en 1 yl ester, and Methyl abietate. Citral, Estragole, Isoborneol, Linalool, Neral, Citronellol, (E,E,Z 1,3,12) Nonadecatriene (5,14 diol), and Methyl eugenol interacted through hydrophobic interactions along with one hydrogen bond. α terpineol, (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester, Bornyl acetate, Diethyl phthalate, Geranic acid, and Rosefuran oxide interacted through hydrophobic interactions along with two hydrogen bonds. Shogaol represents hydrophobic interactions along with three hydrogen bonds; Rosmanol represents hydrophobic interactions along with four hydrogen bonds; Benzenesulfonamide 4 methyl N (5 nitro 2 pyridinyl) and Rosmarinic acid represents hydrophobic interactions along with five hydrogen bonds.Figure 4Two-dimensional plot showing how VP1 protein interacted with different ligands: (i). β pinene, (ii). Camphor, (iii). Caryophyllene, (iv). Citral, (v). Eocalyptol, (vi). Estragole, (vii). Isoborneol, (viii). Linalool, (ix). Neral, (x). n hexadecanoic acid, (xi). α terpineol, (xii). γ terpinene, (xiii). (4 methoxy benzyl) phenethyl amine, (xiv). (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester, (xv). 2,6,10 dodecatrien 1 ol, (3,7,11 trimethyl), (xvi). Benzenesulfonamide 4 methyl N (5 nitro 2 pyridinyl), (xvii). Bornyl acetate, (xviii). Carnosic acid, (xix). Citronellol, (xx). Diethyl phthalate, (xxi). D Limonene, (xxii). (E,E,Z 1,3,12) Nonadecatriene (5,14 diol), (xxiii). Geranic acid, (xxiv). Hydratropic acid undec 2 en 1 yl ester, (xxv). Methyl abietate, (xxvi). Methyl eugenol, (xxvii). Rosefuran oxide, (xxviii). Rosmanol, (xxix). Rosmarinic acid, and (xxx). Shogaol.Selection of one lead compound and its comparative analysis with selected drugsAmong the selected ligands, a lead bioactive natural compound was selected based on several criteria such as the protein–ligand binding predicted as a vina score, drug likeliness properties, and how the ligand and target protein interacted physiochemically.All eight ligands that showed a vina score less than − 8 (Benzenesulfonamide 4 methyl N (5 nitro 2 pyridinyl), Rosmarinic acid, (4 methoxy benzyl) phenethyl amine, Shogaol, Methyl abietate, Hydratropic acid undec 2 en 1 yl ester, (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester, and 2,6,10 dodecatrien 1 ol, (3,7,11 trimethyl)) were selected (Table 3). The subsequent selection of the possible molecule comprises considering factors including the ability to interact with a minimum of four amino acids implicated in subjecting the acceptor substrate and drug-likeliness properties to ADMET profiling (Tables S2–S6). Although there are several properties defined in the ADMET profiles, the major ones that are required for an oral drug were selected. Among the selected ligands, (4 methoxy benzyl) phenethyl amine, (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester, 2,6,10 dodecatrien 1 ol, (3,7,11 trimethyl), Hydratropic acid undec 2 en 1 yl ester, Shogaol, and Methyl abietate showed more than 90% human intestinal absorption while Benzenesulfonamide 4 methyl N (5 nitro 2 pyridinyl), and Rosmarinic acid showed only 81.6% and 32.5% absorption, respectively; therefore, the ligands with > 90% intestinal absorption ability were selected for subsequent selection. The VDss value > 0.45 indicates the good distribution of a drug in tissues, and (4 methoxy benzyl) phenethyl amine showed the highest value among all the selected ligands. (4 methoxy benzyl) phenethyl amine, (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester, 2,6,10 dodecatrien 1 ol, (3,7,11 trimethyl), Hydratropic acid undec 2 en 1 yl ester, and Methyl abietate were predicted to have a blood–brain barrier (BBB) permeability (logBB value) > 0.3; the remaining ligand had a value < 0.3, where logBB value > 0.3 means that compounds can readily cross the BBB. The central nervous system (CNS) permeability (logPS value) determined for (4 methoxy benzyl) phenethyl amine, (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester, 2,6,10 dodecatrien 1 ol, (3,7,11 trimethyl), Hydratropic acid undec 2 en 1 yl ester, Methyl abietate, and Shogaol were >  − 2, whereas the remaining two ligands had a value <  − 2, where a logPS value >  − 2 indicates permeability in CNS. Among the ligands that were found to be permeable to BBB and CNS, (4 methoxy benzyl) phenethyl amine and (Z,Z,Z 6,9,15) octadecatrienoic acid methyl ester showed the best results. But, in further analysis (Z,Z,Z 6,9,15), octadecatrienoic acid methyl ester was found slightly toxic in comparison to (4 methoxy benzyl) phenethyl amine. Therefore, from the above analysis (4 methoxy benzyl) phenethyl amine was selected as the lead compound because of its non-toxic behavior in addition to good intestinal absorption, highest VDss value among the selected ligands, the ability to cross the BBB, and CNS efficiently. Finally, the molecular docking results also put their weight towards (4 methoxy benzyl) phenethyl amine with a vina score of − 8.9.As we already know, no drug has been approved against RVs even though several in vivo and in vitro attempts have been made. Pleconaril is a low molecular weight compound, and the combined analysis of two randomized placebo-controlled trials including individuals that had naturally occurring picornavirus infections showed a shorter duration of illness; meanwhile, the active treatment arm had higher side effects48,49. The Food and Drug Administration (FDA) did not approve the drug due to the side effects and other hindrances such as a reduction in the efficacy of oral contraception taken by volunteers. Although Pleconaril was found initially to be an effective drug, the reason for its rejection was mainly due to its side effects. Nitazoxanide is an oral antiviral drug that targets both enteroviruses and rhinoviruses. Nitazoxanide is currently undergoing clinical development. We thought that the rosemary natural compounds could be an alternative to chemically synthesized drugs; therefore, we performed a comparative study between antiviral drugs, namely, Pleconaril, Nitazoxanide, and (4 methoxy benzyl) phenethyl amine.The drug-likeliness properties of the antiviral drugs Pleconaril and Nitazoxanide were tested and compared against the lead compound (4 methoxy benzyl) phenethyl amine. Although both drugs and the selected compound follow Lipinski’s rule of five, the lead compound was found to be better in terms of molecular weight (Table 2).The ADMET analysisThe water solubility and intestinal absorption of Pleconaril and Nitazoxanide were better than (4 methoxy benzyl) phenethyl amine; meanwhile, in the Caco-2 permeability model, which is another indicator of intestinal absorption, the (4 methoxy benzyl) phenethyl amine was found better. A value higher than 0.9 is considered good in the Caco-2 model. Both the drugs and selected compounds showed skin permeability (logKp) value < −2.5, which is an indicator required to pass in this model. The P-glycoprotein models or ABC transporter models are another indicator for predicting the absorption capability of any drug. Pleconaril was found to be a substrate for P-glycoprotein but an inhibitor for both P-glycoprotein-I and P-glycoprotein-II; on the other hand, the rosemary compound and Nitazoxanide were neither found as a substrate nor inhibitor for P-glycoprotein models (Table S2).The distribution properties were compared based on four predefined models, wherein the VDss model predicted the uniform distribution of any drug, the values above 0.45 show maximum distribution in the tissues, and (4 methoxy benzyl) phenethyl amine was found with good tissue distribution properties; meanwhile, Pleconaril and Nitazoxanide did not exhibit good tissue distribution properties. Furthermore, the values for BBB (Log BB) and CNS (LogPS) permeability were compared, where >0.3 Log BB and > −2 LogPS are considered to cross the BBB and CNS, respectively. Here, (4 methoxy benzyl) phenethyl amine was found to be efficient in crossing these barriers in comparison to Pleconaril and Nitazoxanide (Table S3).The metabolism of any drug involves the most important liver enzyme cytochrome P450 (CYP450). CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 are the major models and are responsible for >90% metabolism of drugs during phase I metabolism. Both drugs and the selected compound showed mixed responses when comparing metabolic properties. (4 methoxy benzyl) phenethyl amine and Pleconaril were found to be substrates for CYP3A4 but not inhibitors of CYP3A4; meanwhile, Nitazoxanide showed the results just opposite to it (Table S4).The excretion property of (4 methoxy benzyl) phenethyl amine was found to be much better than Pleconaril and Nitazoxanide in terms of total clearance value, while only the rosemary compound was found to be a substrate for the renal OCT2 transporter (Table S5).The comparison of predicted toxicity between the lead compound and selected drugs concluded that the maximum tolerated dose of (4 methoxy benzyl) phenethyl amine and Nitazoxanide was higher than the standard value of 0.477; meanwhile, Pleconaril had a dose value in the very low range in this model. Both drugs and the lead compound were not found to be the inhibitors of hERG-I and hERG-II. Oral rat acute toxicity and oral rat chronic toxicity for both drugs and the selected compound were almost in an equal and safe range. The hepatotoxicity model predicted that Pleconaril and Nitazoxanide may cause toxicity and could damage the liver; on the other side, the natural compound, (4 methoxy benzyl) phenethyl amine, was not predicated as a hepatotoxic. Skin sensitization is another model used to predict the future use of the compound as a dermal drug or product, and both drugs and the lead compound were found to not be sensitive to skin. T. pyriformis and minnow toxicity were the models used to detect the drug’s safety towards the environment, where a T. pyriformis toxicity value > −0.5 and minnow toxicity > −0.3 are considered to be safe for the environment; moreover, based on the T. pyriformis model, (4 methoxy benzyl) phenethyl amine, Pleconaril, and Nitazoxanide were predicted as safe while including rhinovirus, as per the minnow model only, (4 methoxy benzyl) phenethyl amine was found as toxic (Table S6).The docking results comparisonThe molecular docking result comparison between the lead compound and selected drugs showed that only (4 methoxy benzyl) phenethyl amine and Pleconaril can target RVs’ VP1 protein with vina scores of −8.9 and −9.0, respectively (Table 3), while Nitazoxanide was found to not be as effective (vina score of −7.0). The similar vina score between the drug and natural compound shows that the efficiency of the natural compound is almost the same as a synthetic drug.