Corrigendum to: Valencene derived from essential oils of Psidium guajava L. as multi-target neurodegenerative inhibitor: a computational study

GC-MS analysis

The analysis of the GC-MS chromatogram revealed 42 peaks in the oil of P. guajava, as shown in the Supplementary material (Fig. S1). It confirmed the presence of 42 compounds, with limonene being the most prevalent one, constituting 30.15% of the area percentage. The mass spectra of each phytochemical, along with its structure, are depicted in the Supplementary Fig. S2 and S3. The name, peak number, molecular weight, area percentage, molecular formula and retention time of the obtained compounds are shown in Table 1.

Binding affinity from molecular docking calculations

The molecular docking calculation was used to assess the best binding pose of the ligands within the active site of the receptor protein, in terms of binding affinity.¹⁹ The results of molecular docking with the MAOB protein revealed β-bisabolene, nootkatene, allo-aromadendrene epoxide, α-copaene and Valencene as the top five best ligands, with binding affinities ranging from −9.993 to −9.794 kcal mol^–1, signifying stronger interaction and binding with the receptor. The highest binding affinity was demonstrated by β-bisabolene with −9.993 kcal mol^–1. The binding affinity of Valencene was −9.794 kcal mol^–1, greater than that of the reference drug Selegiline (−8.817 kcal mol^–1) with the target MAOB.

From the molecular docking calculation with the HTT protein, the best binding affinity was observed with β-bisabolene with a binding affinity of −9.776 kcal mol^–1, followed by 9-epi-(E)-caryophyllene and allo-aromadendrene epoxide, with −9.691 and −9.666 kcal mol^–1 respectively. Similarly, the binding affinity of −9.541 kcal mol^–1 was observed for Valencene. The reference drug Ciproxifan scored binding affinity −10.034 kcal mol^–1, which was found to be slightly higher than that of selected ligands in terms of affinity towards HTT protein.

In the case of the AChE protein, the outcome of molecular docking demonstrated the best binding affinity of −9.559 kcal mol^–1 for guaiac acetate. The binding affinities of −9.526 and −9.305 kcal mol^–1 were obtained for Z-nerolidyl acetate and β-bisabolene respectively. Likewise, the binding affinity of −9.246 kcal mol^–1 was shown by Valencene, which was found to be comparatively better than that of the reference drug, Rivastigmine (−8.883 kcal mol^–1).

Among the studied compounds, the molecular docking of β-bisabolene, allo-aromadendrene epoxide, Valencene (Fig. 1) and nootkatene exhibited better binding affinities with all the neurodegenerative receptor proteins MAOB, huntingtin and acetylcholinesterase, as shown in Table 2 and Supplementary Table S1.

Fig. 1.

Molecular structure of Valencene molecule.

Protein–ligand interactions

The three-dimensional (3-D) and two-dimensional (2-D) representations of the interactions between the ligand and the amino acid residue of the proteins are shown in Fig. 2. It depicts the presence of only hydrophobic interactions, like pi–alkyl, alkyl, pi–sigma and van der Waals between the ligand and receptor proteins, as shown in Table 3. In the interaction of Valencene with the MAOB (4A79) protein, pi–alkyl interactions were observed with the amino acid residues PHE101, TRP117 and TYR324. Likewise, PRO102, LEU162, ILE197 and ILE 314 showed alkyl interaction with the ligand. In the Valencene–HTT (7F61) protein complex, the amino acid residues TYR67, TRP83, TYR88, TYR162, PHE166, TYR262, PHE266 and TRP270 exhibited pi–alkyl interactions with the ligand. The alkyl and pi–sigma interaction was observed with amino acid residues LEU84 and PHE266 respectively. For the hydrophobic interaction, pi–alkyl interactions were observed between the Valencene and the amino acid residues TRP83, TYR238, PHE329, TYR332 and HIS438 of the AChE (7E3H) protein. Several van der Waals interactions were observed between the ligand (Valencene) and different receptor proteins. The interaction between the Valencene molecule and the amino acid residues of the proteins depicted a strong hydrophobic interaction, inferring its potential to inhibit the normal functioning of acetylcholinesterase, monoamine oxidase B and huntingtin proteins.

Fig. 2.

3-D (left) and 2-D (right) representations of the docked ligand at the binding site of the (a) MAOB (4A79), (b) HTT (7F61) and (c) AChE (7E3H) proteins. [Readers should note that the layout of this figure has changed.]

Molecular dynamics simulation (MDS) analysis

MD simulation of the top ligands from docking calculations and reference drugs complexed with different target proteins was performed to analyse the stability of the adduct formed. The geometrical stability of ligands in their adducts can be assessed through the root mean square deviation (RMSD) profiling obtained from the MDS trajectory.²⁰ The RMSD of the protein backbone and ligand with respect to the protein backbone depicts the overall stability of the formed adducts as shown in Fig. 3.²¹ Among top-docked ligands, the molecular dynamics simulation of Valencene showed a good stability with proteins (PDB ID: 4A79, 7F61 and 7E3H), with RMSD below 6.5 Å. The RMSD plots of the other three ligands, with respect to the protein backbone, were found to be inconsistent throughout the production run and are presented in Supplementary Fig. S4 and S5. The RMSD trajectories of aromadendrin (Fig. S4) and bisabolene (Fig. S5) with different target proteins were found to be reasonably unstable with high RMSD values. The reasonably smooth RMSD curve of the ligand demonstrated the stability of the adduct during the production run in terms of the conservation of the docked pose and position. An exceptional stability was shown by Valencene with the MAOB protein, with a consistently smoother RMSD trajectory of ~4 Å despite some initial spikes (Fig. 3a). The ligand RMSD of ~3.0 Å, with slight fluctuation and spikes, was depicted by Valencene with the AChE protein, implying that the ligand remained docked at the same location throughout the production run (Fig. 3b). The Valencene exhibited moderate stability with the HTT protein, with a RMSD of ~6.0 Å, along with some fluctuations (Fig. 3c). Despite major fluctuations and spikes in the trajectory, the system attained equilibrium after 70 ns. Although the reference Selegiline was found to be stable after 30 ns with a RMSD of ~2.5 Å in complexation with MAOB protein (Fig. 3d), Valencene depicted a more stable trajectory at 4 Å with the same target. Ligand RMSD of another reference drug Rivastigmine complexed with AChE, showed a gradually increasing trajectory, initiating from 4 Å up to 100-ns production run (Fig. 3e), which was found to be comparable to the trajectory formed by ligand RMSD in the Valencene–AChE complex. Likewise, the Ligand RMSD trajectory in the HTT–Ciproxifan complex was found to be stable at 5 Å with few spikes at initiation and end of the production run (Fig. 3F). So, the ligand RMSDs of the Valencene with multiprotein clearly showed their comparable geometric stability as found in various reference drugs targeting three neurodegenerative proteins. A smooth curve with a RMSD of ~2.0 Å of the protein backbone inferred the stability of protein structure upon ligand binding. No significant structural changes were observed.

Fig. 3.

RMSD of Valencene, protein backbone and reference drugs (Selegiline, Rivastigmine and Ciproxifan) with respect to protein backbone in different adducts (a) MAOB (orange)–Valencene (black), (b) AChE (magenta)–Valencene (black), (c) HTT (indigo)–Valencene (black), (d) MAOB (brown)–Selegiline (cyan), (e) AChE (red)–Rivastigimine (green) and (f) HTT (maroon)–Ciproxifan (blue). [Readers should note that the layout of this figure has changed.]

The root mean square fluctuation (RMSF) depicts the extent of fluctuation and flexibility of the amino acid residues of a protein over the 100-ns production run. Higher RMSF indicates greater flexibility during simulation, whereas lower RMSF indicates minimal changes, implying a tightly bound protein–ligand complex and enhanced system stability.²² The RMSF plot (Fig. 4a) shows that the fluctuation of α-carbon atoms of the MAOB protein was less than 2.0 Å for almost all amino acid residues except for the terminal part. Similarly, the RMSF of the amino acid residue of the AChE and HTT proteins were less than 3.0 Å, with an exception at ~140 and 155 amino acid residue numbers of the 7F61 protein. The RMSF trajectory of AChE and HTT proteins with Valencene and their respective reference drugs (Rivastigmine and Ciproxifan) are observed to be comparatively similar (Fig. 4). Surprisingly, RMS fluctuation of protein α-carbon atoms in the MAOB–Valencene complex was observed to be comparatively higher than found in the MAOB–Selegiline complex (Fig. 3d).

Fig. 4.

RMSF plots of protein backbones in complexes (a) MAOB–Valencene (orange), (b) AChE–Valencene (magenta), (c) HTT–Valencene (indigo), (d) MAOB–Selegiline (Brown), (e) AChE–Rivastigimine (red) and (f) HTT–Ciproxifan (maroon). [Readers should note that the layout of this figure has changed.]

The radius of gyration (R_g) obtained through MDS reflects the compactness of protein–ligand complexes, with a smaller R_g suggesting a more compact structure.²³ From the calculation of R_g, a steady and smooth trajectory at ~23 ± 0.7 Å indicated no significant expansion or shrinkage of the proteins, implying the stability of all three proteins (4A79, 7E3H and 7F61) upon the binding of the ligand as evident from Fig. 5. The R_g of protein obtained for AChE (Fig. 5b) and HTT (Fig. 5e) docked with reference drugs were found to be similar to those found for the docked complex with Valencene at ~23 and ~22 Å respectively. The R_g plot of the MAOB protein docked with the drug Selegiline was found to have various spikes throughout the MDS run (Fig. 5d), but the MAOB complexed with Valencene was comparatively smooth. Such an undatable R_g trajectory of the protein with Selegiline indicates a more stable complex formation of MAOB–Valencene than the Selegiline–MAOB complex.

Fig. 5.

R_g of protein in complexes (a) MAOB–Valencene (orange), (b) AChE–Valencene (magenta), (c) HTT–Valencene (indigo), (d) MAOB–Selegiline (Brown), (e) AChE–Rivastigimine (red) and (f) HTT–Ciproxifan (maroon). [Readers should note that the layout of this figure has changed.]

The solvent-accessible surface areas (SASAs) refer to the area of a protein’s surface that interacts with its solvent molecules.²⁴ A reasonably smooth trajectory with SASAs of 216 ± 4 nm² was observed with MAOB–Valencene and AChE–Valencene complexes. Likewise, a smooth SASA curve with a value of ~175 nm² was obtained for the HTT–Valencene complex (Fig. 6). The SASAs remained almost stable throughout the production run without any significant bumps and spikes, indicating no change in the wettable area of the protein upon ligand binding over time in the Valencene complexes. Though few initial bumps were noticed in MAOB (Fig. 6d) and HTT (Fig. 6f) complexed with the reference drug, the SASA of all the reference drugs was observed to be similar to that of the complex formed by Valencene with different proteins, which is presented in Fig. 6. The combined analysis of all the geometric parameters, RMSD, SASA, RMSF and R_g demonstrated the good stability of the different target protein structures in complexation with Valencene and respective reference drugs. The comparative stability assessment of three distinct neurodegenerative target proteins bound to Valencene and reference drugs further confirmed the strong modulatory potential of Valencene in comparison to the reference drugs.

Fig. 6.

SASAs of protein in complexes (a) MAOB–Valencene (orange), (b) AChE–Valencene (magenta), (c) HTT–Valencene (indigo), (d) MAOB–Selegiline (Brown), (e) AChE–Rivastigimine (red) and (f) HTT–Ciproxifan (maroon). [Readers should note that the layout of this figure has changed.]

Binding free energy changes (ΔG_BFE) or thermodynamic stability

The binding free energy change (ΔG_BFE) was used to assess the feasibility and spontaneity of the complex formation: the lower the negative value of ΔG_BFE, the greater the stability of the protein–ligand system.^25,26 The free energy changes of the adducts calculated for an equilibrated part of the trajectory (20 ns) are shown in Table 4. The best ΔG_BFE was observed for the MAOB–Valencene complex with −25.18 ± 2.30 kcal mol^–1. Similarly, negative binding free energy changes of −15.17 ± 3.44 and −18.98 ± 2.38 kcal mol^–1 were observed for the AChE–Valencene and HTT–Valencene complexes respectively. ΔG_BFE of the MAOB–Valencene and HTT–Valencene complexes were detected comparatively greater in value than that of MAOB–Selegiline and HTT–Ciproxifan complexes respectively. The result supported the better stability and spontaneity of the complex formation reaction of Valencene with MAOB and HTT proteins compared to those of the reference drugs (Table 4). However, the AChE–Rivastigmine complex was found to be formed with the release of slightly higher energy than in the formation of the AChE–Valencene complex. So, ΔG_BFE for all three complexes suggested the spontaneous nature of complex formation and a more favourable forward complex formation reaction of Valencene with MAOB and HTT proteins compared to those of the complexes of the reference drugs.

Drug-likeness and toxicity

The pharmacokinetic and pharmacodynamic characteristics of Valencene were predicted by in silico tools to evaluate its drug-likeness and suitability for human consumption, as depicted in Table 5. From the prediction of absorption, metabolism, excretion, distribution and toxicity (ADMET), it was observed that the Valencene molecule belongs to toxicity class 5 and obeys the Lipinski’s rule of 5 (RO5), suggesting a probability for drug-likeness. The compound did not show any sign of immunotoxicity, carcinogenicity, mutagenicity, cytotoxicity or hepatotoxicity but exhibited skin sensitisation and neurotoxicity, analogous to the reference drugs, selegiline and rivastigmine. The blood–brain barrier (BBB) plays a very crucial role in the treatment of neurodegenerative disorders, as a majority of molecules encounter difficulties in crossing this barrier.²⁷ The hit candidate and both reference drugs demonstrated the potential to cross the BBB, indicating their capability to function as brain-targeting drugs. The metabolic activity of Valencene was similar to that of the drugs and exhibited no hERG-blocking properties, implying that the compound does not induce fatal ventricular arrhythmia. The compound demonstrated high intestinal absorption, like the reference drugs, representing effective passage of the compound through the intestinal cell membranes suitable as an oral formulation.²⁸ The total clearance of the compound was almost similar to the clearance of the reference drug, suggesting its potential to be used as a drug.

The comparative evaluation of toxicity and pharmacokinetics of the Valencene compound with reference drugs revealed its less toxic nature along with its potential to treat neurodegenerative disorders. The ADMET results suggested further exploration of the hit compound through experimental trials, in vivo and in vitro, to assess its suitability for human consumption.

Collection of plant samples and extraction of essential oil

The leaves of Psidium guajava L. were collected from Kathmandu, Nepal, and verified by the National Herbarium and Plant Laboratories, Lalitpur, Nepal. The essential oil from the leaves was extracted through hydrodistillation using a Clevenger-type apparatus.²⁹

GC-MS analysis

Gas chromatography–mass spectrometry (GC-MS) experiments were performed using a GCMS-QP 2010 instrument. Helium was used as the carrier gas, passing through an RTX-5MS column of dimension 60 m × 0.32 mm × 0.25 μm. The temperature profile included a gradual increase from 80 to 300°C, with respective hold times for the start and end of 2.0 and 5.0 min. Throughout the analysis, the ion source and interface were consistently maintained at 200 and 250°C respectively, and the compound identification relied on MS comparison using the Flavor and Fragrance Natural and Synthetic Compounds (FFNSC, ver. 4.0) library.³⁰

In silico approach

Molecular docking calculations

The molecular docking calculation was carried out using the DockThor server (see https://dockthor.lncc.br/) to determine the best binding pose of the ligand within the receptor protein.³⁵ The docking program was selected because of its capability to perform numerous docking calculations quickly, being freely available and for its ability to reproduce the results. Different control parameters were used for the docking of ligands with different proteins. For the protein 4A79, box coordinates (51, 157, 31), box size (15, 15, 15), discretisation (0.16), population size (750), numbers of evaluations (1,000,000) and numbers of runs (24) were selected. The box size (16, 16, 16), box coordinates (−43, 36, −32), discretisation (0.17), population size (750), number of runs (24) and evaluations (1,000,000) were selected for protein 7E3H. Similarly, box coordinates (−19, 50, −1), box size (15, 15, 15), discretisation (0.16), population size (750), number of runs (24) and evaluations (1,000,000) were used for protein 7F61. The validation of the docking protocol was done by obtaining the RMSD of less than 2 Å for each (1.420, 0.669 and 1.666 Å for MAOB: 4A79, AChE: 7E3H, and HTT: 7F61 proteins respectively) through the superimposition of co-crystallised ligands in the crystalline complexes with re-docked ligands as shown in Fig. 7.

Fig. 7.

Superimposition of native (cyan) and re-docked ligand (red) in (a) MAOB (PDB ID: 4A79), (b) HTT (PDB ID: 7F61), and (c) AChE (PDB ID: 7E3H).