Computational studies on the cholinesterase, beta-secretase 1 (BACE1) and monoamine oxidase (MAO) inhibitory activities of endophytes-derived compounds: towards discovery of novel neurotherapeutics

Abstract Cholinesterases, beta-secretase 1 (BACE1) and monoamine oxidase (MAO) are significant in the etiology of neurodegenerative diseases. Inhibition of these enzymes is therefore a major strategy for the development of neurotherapeutics. Even though, this strategy has birthed some approved synthetic drugs, they are characterized by adverse effects. It is therefore, imperative to explore promising alternatives. Consequently, we assessed the inhibitory activities of some endophytes-derived compounds against selected targets towards discovery of novel neurotherapeutics. Standard inhibitors and 83 endophytes-derived compounds were docked against acetylcholinesterase (AChE), butyrylcholinesterase (BChE), BACE 1 and MAO using AutodockVina while the molecular interactions between the selected targets and the compounds with notable binding affinity were viewed through Discovery Studio Visualizer. Druglikeness and Absorption–Distribution–Metabolism–Excretion-Toxicity (ADMET) and blood brain barrier (BBB) properties of the top 4 compounds were evaluated using the Swiss online ADME web tool and OSIRIS server; ligands-enzymes complex stability was assessed through molecular dynamics (MD) simulation. From the 83 compounds, asperflavin, ascomfurans C, camptothecine and corynesidone A exhibited remarkable inhibitory activity against all the four target enzymes compared to the respective standard inhibitors. However, only corynesidone A could transverse the BBB and predicted to be safe. MD simulation of the unbound and complexed enzymes with corynesidone A showed that the complexes were stable throughout the simulation time. Given the exceptional inhibitory activity of endophytes-derived corynesidone A against the four selected targets, its ability to permeate the BBB, excellent drugability properties as well as its stability when complexed with the enzymes, it is a good candidate for further studies towards development of new neurotherapeutics. Communicated by Ramaswamy H. Sarma


Introduction
Neurodegenerative disorders include a spectrum of diseases typified by gradual and irremediable loss of neurons, neurotransmitters and neuronal structures from particular regions of the brain. This spectrum include Parkinson disease (PD) and Huntington's disease (HD), where loss of neurons and neuronal structures of the basal ganglia leads to anomaly in the coordination of movement; Alzheimer's disease (AD), where the loss of hippocampal and cortical neurons and neuronal structures causes memory loss and cognitive deficiency; and amyotrophic lateral sclerosis (ALS), where muscular weakness results from the degeneration of spinal, bulbar, and cortical motor neurons (Harms et al., 2018). There is also a buildup of abnormal proteins in the brain tissues (Dugger & Dickson, 2017). These disorders are relatively common and constitute a substantial medical and public health concern in all human populations. Among the current treatment strategies for the treatment of neurodegenerative diseases is the reversal of neuron, neurochemicals and neuronal structures that are lost during the pathogenesis of the disease process.
PD is characterized by loss of dopamine and dopaminergic neuron and one of the strategies employed to retain the levels of dopamine in the substantial nigra pars compacta (Surmeier, 2018) is the inhibition of mono amine oxidase (MAO), which catalyses oxidative deamination of dopamine and other biogenic amines. It is plausible that inhibition of MAO would increase the levels of dopamine in the stratum and alleviate some of the symptoms of PD and other neurodegenerative diseases (Finberg & Rabey, 2016;Behl et al., 2021).
The hallmark of neurochemical disturbances and imbalance in AD is the deficiency of acetylcholine, therefore, augmentation of the cholinergic transmission is currently the basis of AD treatment (Hampel et al., 2018). This involves the use of reversible inhibitors of cholinesterases, enzymes that act to limit choline neurotransmission by catalyzing the breakdown of acetycholine to choline and acetate in the synaptic cleft. Acetylcholinesterase is the major cholinesterase in the brain while butrylcholinesterase is a serum and hepatic cholinesterase that is upregulated in AD brain. Cholinesterase inhibitors (ChEIs) are usually the first line therapy for cognitive decline in AD. They are also widely used to treat other neurodegenerative disorders associated with low levels of cholinergic transmission such as lewy bodies and dementia (Jellinger & Korczyn, 2018).
Furthermore, beta-secretase (also known as B-site amyloid precursor protein cleavage enzyme-BACE1), a type 1 integral membrane glycoprotein, has also been implicated in the development of AD (Das & Yan, 2019). Decades of research into the cause of AD have linked the increase in the incidence of AD to the accumulation of beta amyloid (A-beta) peptide in the brain. A-beta is a sticky compound, whose aggregation in the brain disrupts neural communication and structures with attendant neuro-degradation and cell death (Decourt et al., 2021). Since the production of A-beta is by proteolytic cleavage of amyloid precursor protein (APP) by BACE1. It is therefore, conceivable that the inhibition of BACE1 activity provides a likely therapeutic target for the treatment of AD. These lines of research have been actively and vigorously pursued by scientists and the research findings had led to the production of certain drugs which are currently used in the treatment of neurodegenerative diseases. However, most of these drugs have side effects and lesser efficacies due to factor such as low bioavailability, polarity, ability to cross the blood brain barrier (BBB) and other factors listed in the Lipinski's rule of five for drug availability. Therefore, there is an increasing search for novel neurodegenerative therapeutics with more efficacy, lesser side effects and better pharmacokinetics parameters.
Endophytes are interesting group of microbes with the ability to produce a broad range of secondary metabolites characterized by significant pharmaceutical potentials (Kaushik & Coloma, 2020;Praptiwi et al., 2018). Some of the reported therapeutic properties exhibited by endophytes include antioxidant, antiinflammatory, antimicrobial, anticancer, antidiabetic, antiviral, neuroprotective, and hepatoprotective properties (Falade et al., 2021). The bioactivity of endophytes has been attributed to the presence of a wide range of compounds belonging to alkaloid, xanthones, depsipeptide, bicyclic lactones, depsidones, butenolides, ergosterol, spirobisnaphthalenes, benzopyran derivatives, isofuranonaphthalenone, sesquiterpenoids, pestalols etc (Falade et al., 2021). Undoubtedly, endophytes-derived compounds hold promising potentials for drug development in the management of various diseases. Therefore, this study aimed at evaluating the inhibitory properties of endophytes-derived compounds against selected drug targets (AChE, BChE, BACE1 and MAO) in a quest to finding novel neurotherapeutics using in silico methods. In silico methods such as molecular docking and molecular dynamics are generally-accepted computational tools employed to investigate molecular recognition to predict the binding affinity and mode of complexes formed by ligands and proteins (Ishola & Adewole, 2019b). For example Singh et al. (2020) used these computational methods including docking and molecular dynamics to identify an aminoarylbenzosuberene as a potential inhibitor of checkpoint kinase 1 with potential implication in the treatment of cancer. In other studies, in silico methods have been used to identify some natural molecules as c-aminobutyric acid receptor antagonist , pyrrolone-fused benzosuberene compounds as cyclin-dependent kinase 2 inhibitors  and dual specificity tyrosine-phosphorylation-regulated kinase inhibitors .

Mining of endophytes-derived compounds
Endophytes related research articles were retrieved from web of science, scopus and google scholar databases as reported by Falade et al. (2021). About 246 endophytes-derived compounds were compiled (Supplementary List 1). However, 83 compounds were used for this study based on availability of the chemical structures.

Protein preparation
The crystal structures of acetylcholinesterase (AChE) and butyrylcholinesterase (BChE), beta-secretase 1 (BACE1), monoamine oxidase (MAO) with PDB IDs 6O4W (Gerlits et al., 2019), 4B0P (Wandhammer et al., 2013), 6EJ3 (Johansson et al., 2018), 1GOS (Binda et al., 2002), respectively were retrieved from the protein databank (http://www.rcsb.org). The structures were prepared individually by eliminating existing ligands and water molecules, while the absent hydrogen atoms were added using the Autodock v4.2 program, Scripps Research Institute. Polar hydrogen charges of the Gasteiger type were allocated and the nonpolar hydrogens were integrated with the carbons and the internal degrees of freedom and torsion were formed. Target proteins were subsequently saved into pdbqt format in preparation for molecular docking.

Ligand preparation
The SDF structures of donepezil, elenbecestat, isocarboxazid and 83 other compounds were obtained from the PubChem database (www.pubchem.ncbi.nlm.nih.gov). Using Open babel program (O'Boyle et al., 2011), the compounds were converted from SDF to mol2 chemical format. The ligand's alpha carbon was detected when the internal degrees of freedom and torsion were set to zero. Further, the compounds were converted using Autodock tools to a dockable pdbqt format.

Molecular docking
Docking of the compounds to AChE, BChE, BACE 1 and MAO as well as the assessment of binding affinities were done by using Vina GUI (Trott & Olson, 2010). The pdbqt format of the proteins and the ligands were dragged into their respective columns. The grid center for docking was detected as X ¼ 91. Subsequently, the program was run and an exhaustive series of docking calculations across the whole protein surface using a gradient optimization algorithm in order to find the spots with best binding affinities. A cluster analysis based on Root Mean Square Deviation (RMSD) values for starting geometry was conducted and the lowest energy conformation of the more populated cluster was found to be the most accurate solution. The pose with the strongest affinity for each cluster was taken as the representation of this cluster. The compounds were then ranked by their affinity scores. Thereafter, molecular interactions between the selected targets and the compounds that have the highest binding affinity were viewed with Discovery Studio Visualizer, 2020.

ADMET and BBB properties prediction
The ADMET properties of the three standard inhibitors used in this study (donepezil, elenbecestat, isocarboxazid) and four compounds with notable binding affinity for the four proteins studied were evaluated. ADME studies were carried out using the Swiss online ADME web tool (Daina et al., 2014(Daina et al., , 2017Daina & Zoete, 2016) to evaluate the druglikeness of the phytochemicals that were selective for the three protein drug targets while the US Food and drug administration toxicity risk predictor tool OSIRIS evaluated various toxicity risks properties such as tumorigenicity, mutagenicity, irritation, and reproductive development toxicity. A graph of WLOGP against TPSA was plotted using GraphPad Prism 6 software (GraphPad Software, California, USA) to determine the BBB properties of the compounds as described by (Ishola & Adewole, 2019a).

Molecular dynamics
The enzymes (AChE, BChE, BACE 1 and MAO) complexed with corynesidone A and the unbound enzyme were subjected to full atomistic Molecular Dynamic (MD) simulation using GROMACS 2019.2 and GROMOS96 43a1 forcefield (Abraham et al., 2015;Bekker et al., 1993;Oostenbrink et al., 2004). Corynesidone A topology files were generated using PRODRG webserver (http://davapc1.bioch.dundee.ac.uk/cgibin/prodrg) (Sch€ uttelkopf & Van Aalten, 2004). The enzymes and corynesidone A-enzymes complex systems were solvated within a cubic box of the transferable intermolecular potential with a four-point (TIP4P) water model, with periodic boundary conditions applied at a physiological concentration of 0.154 M set by neutralized NaCl ions. The number of water molecules, Na þ and Clions for the four enzyme systems are AChE: 82,436,63,71;BChE: 78,108,62,60;BACE 1: 61,408,47,55 and MAO: 115,356,86,89, respectively. The minimization of the systems was performed for 10,000 steps using steepest descent algorithm in constant number of atoms, volume, and temperature ensemble (NVT) ensemble for 0.3 nanosecond, followed by 0.3 nanosecond of equilibration in constant number of atoms, constant pressure and constant temperature (NPT). Temperature and pressure were set to 310 K and 1 atm and maintained using velocity rescale and Parrinello-Rahman barostat, respectively. Leap-frog integrator was used with a time step of 2 femtosecond. For each system, 100 ns of simulation were performed and for each 0.1 nano a snapshot was saved with a total of 1000 frame from each system. The trajectories were analyzed using VMD TK console scripts (Humphrey et al., 1996) to calculate Root mean square deviation (RMSD), Root mean square fluctuation (RMSF), Surface accessible surface area (SASA), Radius of gyration (RoG) and number of H-bond. A t-Test: Paired two samples for means were calculated to analysis the level of difference between the apo and bound enzymes using the thermodynamic parameters.

Binding interactions
Results showed that four compounds exhibited notable binding affinity for all the four proteins considered in this study; these include asperflavin, isolated from Eurotiumcristatum EN-220 (Du et al., 2014); ascomfurans C, from Ascomycota sp. SK2YWS-L ; camptothecine, which has been isolated from Fusarium oxysporum kolhapunensis, Colletotrichum fructicola SUK1 and Corynespora cassiicola SUK2 (Bhalkar et al., 2015(Bhalkar et al., , 2016; and corynesidone A, from Corynespora cassiicola (Chomcheon et al., 2009;Okoye et al., 2013a;Zhao et al., 2016). The binding interactions of asperflavin, ascomfurans C, camptothecine are presented as Supplementary Figures 1-8 while the binding interactions between corynesidone A and all the four proteins are shown in Figures 1-4. Figure 1(a) presents the 3D cartoon and stick view of interaction between amino acid in acetylcholinesterase binding sites and donepezil, Figure 1(b) presents the 3D cartoon and stick view of interaction between amino acid in acetylcholinesterase binding sites and corynesidone A. Figure  1(c) presents the interaction between amino acids in the binding site of acetylcholinesterase and donepezil, while Figure 1(d) presents the interaction between amino acids in the binding site of acetylcholinesterase and corynesidone A. From Figure 1(c), results showed that hydrophobic and carbon-hydrogen interactions dominated as the main form of interaction between donepezil and AChE with hydrophobic bonds, via p-sigma bonds with TRP86, TRP286, LEU289 and HIS447, and carbon-hydrogen interaction was observed with ASP74, THR83 TRP86 and SER203, while a single hydrogen bond was established with TYR124. Likewise, hydrophobic was also the main mode of interaction between corynesidone A and AChE, with the ligand interacting with the protein via p-p stacking with TRP86, TYR337 and PHE338, whereas a single carbon-hydrogen bond with HIS447 and a single hydrogen bond with SER125 were also observed. Results show that hydrophobic mode of interaction also predominates between donepezil and BChE via p-p stacking with TRP82 and via p-sigma bonds with ALA277, ALA328, TRP430, MET437 and HIS438. While no hydrogen bond was observed between donepezil and BChE, carbon-hydrogen bonds were seen with ASN68, GLU197, GLY439 and TYR440. Hydrophobic mode of interaction also predominates between corynesidone A and BChE p-p stacking with TRP82 and PHE329 but via p-sigma bond with ALA328, there was a single carbon-hydrogen bond with HIS438 and hydrogen bond with ALA328. For elenbecestat and BACE1, four different types of interactions were observed, including carbonhydrogen, hydrogen bond, hydrophobic bond and halogen bond. Halogen bond was formed with PHE108 and ARG128, hydrophobic bond, via p-sigma bond with TYR71, hydrogen bond with ARG128 and THR231, and carbon-hydrogen bond with GLN73. For corynesidone A and BACE1, hydrophobic interaction occurred via p-sigma bond with VAL69, pp-stacking with TYR71, and hydrogen bond with TRP76. In the case of isocarboxazid and MAO, the major form of interaction was hydrophobic via p-sigma bond with LEU171, CYS172, ILE199, TYR326, TYR398 and TYR435, with a single carbon-hydrogen bond with ILE199 and hydrogen bond with TYR435. Between corynesidone A and MAO, hydrophobic bond via p-sigma bond with LEU171, p-p-stacking with PHE343 and TYR435, with a single hydrogen bond with TYR435 and carbon-hydrogen with GLN206.

ADMET and BBB properties prediction
The top compounds with molecular docking results for the four proteins were submitted to ADMET and BBB properties prediction, in order to evaluate their potential as neuroactive drug agents. We conducted the ADMET analysis and BBB prediction of asperflavin, ascomfurans C, camptothecine and corynesidone A. From the results ( Table 2), none of the four compounds violate the Lipinski's rule of drugability as they all had molecular weights less than 400 g/ml; hydrogen bond acceptor of less than 10; hydrogen bond donorofless than 5 and MLogP value not > 4.5. Furthermore, toxicity evaluation showed that corynesidone A and camptothecine pose no risk as there was no toxic fragment detected as against ascomfurans C, which is predicted to be both tumorigenic and irritant while asperflavin could pose a high irritation risk (Table 2). However, only corynesidone A passed the BBB permeability test as reflected in an 'egg yolk-like model', in which all the BBB non-permeative compounds fell outside the sphere shape, with only corynesidone A within the sphere ( Figure 5).

Molecular dynamics simulation of lead complexes
The structural integrity of the proteins complexed with corynesidone A was studied in a full atomistic MD simulation. The following thermodynamic parameters (RMSD, RMSF, SASA, RoG and number of H-bonds) were computed from the trajectories; the plots were presented as a function of time frame and compared with the apo enzyme as the reference structure. The mean values for the five computed thermodynamic parameters are presented in Table 3. The RMSD plots measure the stability of the protein structure by observing the extent of the deviation from the initial structure. Figure 7 shows the RMSD plots of the four enzymes complexed with corynesidone A and the apo enzymes. The four systems were equilibrated around 4, 6, 10 and ns for the AChE, BChE, BACE 1 and MAO enzyme system, respectively. They maintained a steady progression to the end of the simulation with minimal fluctuations. When compared to the complexed system, the unbound BACE1 enzyme displayed higher fluctuations at 78 ns this was responsible for the higher RMSD mean value (Figure 7c). For each of the enzymes, the apo and complexed system presented close average RMSD values except for MAO with higher mean RMSD values than the unbound enzyme (Figure 7d).
The RMSF plots reveal the flexibility of the amino acid residues of the protein. Higher fluctuations are observed at the N and C terminal ends of the proteins due to terminal motions. Figure 8 shows the RMSF plots of the four enzymes for the apo and complexed with corynesidone A. From the computed average values, the complexed system presented a lower average RMSF values except for MAO. This further indicates that the binding of corynesidone A formed more compacted structures.
The extent of the compactness of protein upon binding of the ligands is measured from the RoG plots and values. A stably folded protein structure presents a steady RoG plot. Figure 9 shows the RoG plots of the four systems. The plots for the systems show a steady progression with minimal fluctuations. The mean RoG values calculated for the four systems showed that the binding of the corynesidone A to the active site of the enzymes did not distort the structure integrity of the enzymes. The lower average RoG values for the complexed system further indicates a more compacted structure upon the binding of the corynesidone A.
The SASA plots and computed mean values measure the solvent accessible by the surface of the enzymes. Both RoG and SASA plots indicate the level of structural unfolding of proteins with reference to its original structure. Figure 10 shows the SASA plots of the four systems. The plots for the unbound and complexed structures for proteins showed similar patterns with minimal fluctuations except for the MAO that presented a higher mean SASA value when compared to the unbound structure. The average number of hydrogen bonds for the unbound and complexed system for the four enzymes is presented in Figure 11. A slight reduction in average number of hydrogen bond was observed in the complexes when compared to the unbound protein (Table 3). The result of the two-tailed T-test for the thermodynamic parameters showed that except for the mean RMSF values of AChE and BACE 1 and mean number of H-Bonds for BChE with t-values of 0.148, 0.7640 and 0.0551, respectively, there is no significant difference (a ¼ 0.05) between the apo and the binding of corynesidone A for all the four enzymes (Table 3).

Discussion
Molecular docking has been widely used in computational drug design to identify new and potent ligands of therapeutic importance and predict the binding mode and affinity of a protein-ligand complex (Bartuzi et al., 2017). Specifically, docking model has been employed for identification of many potential inhibitors of enzymes relevant to different pathologies such as cancer, AD and diabetes mellitus (Ishola & Adewole, 2019b). For instance, Jabir et al. (2021b) screened one hundred associated-glycogen synthase kinase-3b (GSK-3b) ligands for their ability to inhibit AChE through molecular docking and identified thirteen as potential cholinesterase inhibitors, with promising neurotherapeutic potential. Similarly, the identified AChE inhibitors were screened for inhibition of BACE-1 and c-secretase, which are also potential targets for development of neurotherapeutics (Jabir et al., 2021a).
Thus, 4 of the screened ligands (asperflavin, ascomfurans C, camptothecine and corynesidone A) in this study displayed higher or similar negative binding affinities for all the enzyme targets (AChE, BChE, BACE1 and MAO) compared to the respective standard inhibitors: donepezil, elenbecestat and isocarboxazid. The remarkable inhibitory activities of these compounds may be attributed to the presence of a number of hydroxyl groups and oxygen atoms in their structural moiety (Figure 6), which may serve as hydrogen bond donor or hydrogen acceptor during interaction with amino acid residues in the active sites of the target enzymes, thereby enhancing their binding affinities. This study however, suggests corynesidone A as the most promising candidate for development of neurotherapeutic agent due to its predicted ability to transverse the BBB, which was lacking in the other three potential inhibitors: asperflavin, ascomfurans C and camptothecine. Despite this limitation, the enlistment of asperflavin, ascomfurans C and camptothecine as neurotherapeutic agents may be possible by using selective vasoactive agents such as leukotrienes, histamine and bradykinin, to improve brain permeability (Black, 1995;Ishola et al., 2021). More so, BBB could be circumvented by administering the therapeutic agents via an alternative route such as intranasal delivery (ID). Intranasal administration allows BBB non-permeative therapeutic agents to be transported to the central nervous system (CNS) in a short time (Hanson & Frey, 2008). This is achievable due to the special links that the olfactive and trigeminal nerves offer between the brain and the external environment (Hanson & Frey, 2008). In fact, the use of ID would not require structural or any other modification of the neurotherapeutic agents (Hanson & Frey, 2008), hence the potency of the compounds would still be maintained.
Corynesidone A is a depsidone, which was first isolated from the endophytic fungus, C. cassiicola L36 by Chomcheon et al. (2009). Subsequent isolations of corynesidone A were from C. cassiicola hosted by Gongronemalatifolium (Okoye  (Zhao et al., 2016). The endophyte-derived depsidone is characterized by promising pharmaceutical potentials due to its reported aromatase inhibitory, antioxidant and antiinflammatory properties (Chomcheon et al., 2009;Okoye et al., 2013b). Generally, depsidones have shown to have interesting pharmacological properties including antiviral, antibacterial and antioxidant properties (Legaz et al., 2011). However, there is scarcity of data on the neuroprotective potential of this class of compounds. This study is perhaps, the first report on the neurotherapeutic potential of corynesidone A as evident in its multi-target inhibition of AChE, BChE, BACE1 and MAO. This assertion is buttressed by its remarkable antioxidant properties (Chomcheon et al., 2009), which suggest its significance in the management of oxidative stress-induced neurodegenerative diseases. Besides, the anti-inflammatory activity of corynesidone A (Okoye et al., 2013b) positioned it as a potential drug candidate for the treatment of neuro-inflammatory disorders. Also, there are scientific evidences that link inflammatory process with multiple neurodegenerative pathways (Chen et al., 2016).
Lipinski Rule-of-five, which defines the relationship between pharmacokinetic and physicochemical parameters, states that, generally, an orally active drug may not violate more than one of the following criteria: (1) A molecular mass < 400 g/ml (2) Not >10 hydrogen bond acceptors, (3) Not >5 hydrogen bond donors and (4) An octanol-water partition coefficient log P not >5 (Lipinski, 2000). Corynesidone A has all its data within the normal range as specified by Lipinski. Subjecting the compounds to the scrutiny of Verber's rule by means of the 'polar surface area and the number of rotatable bonds' (Veber et al., 2002), corynesidone A satisfied the Veber's rule which also supports its prospect as an oral drug agent. Moderately polar (TPSA < 79 Å2) and relatively lipophilic (WLOGP from 0.4 to 6.0) compounds have a high possibility of access to the CNS (Daina & Zoete, 2016). This is validated by the egg-yolk-like model where corynesidone A aggregated in the sphere just like donepezil and isocarboxazid ( Figure 5). The BBB is seen as a barrier shielding the brain via a 'physical' barricade and a 'biochemical' barricade made of enzymatic actions and dynamic efflux (Daina & Zoete, 2016), thus, serving as a vital wall between the CNS and systemic circulation (Goodwin, 2005). This BBB phenomenon has been a great challenge in CNS drug delivery despite the numerous drug discovery Figure 5. Blood brain barrier properties of donepezil, elenbecestat, isocarboxazid, corynesidone A, asperflavin, ascomfurans C and camptothecine. Three permeant (BBBþ, yellow dot inside black sphere), 4 non permeants (BBB-, orange dots outside black sphere). efforts in both the academia and the industry (Pardridge, 2009). Therefore, the ability of corynesidone A to cross the BBB, coupled with its potential to serve as a multi-target inhibitor of AChE, BChE, BACE1 and MAO, indicate that it is a promising lead compound and may find application as an anti-Alzheimer's disease molecule. These results will also serve a fine starting point for further experimental studies.
The analysis of the plots of the various thermodynamic parameters (RMSD, RMSF, RoG, SAS, and H-bonds number) computed from the trajectories files of the 100 ns MD  simulation of the unbound and complexed enzymes with corynesidone A showed that the complexes were stable throughout the simulation time (Cheng & Ivanov, 2012). The close average-values for each parameter for the complexes when compared to the unbound protein further showed that the binding of corynesidone A did not destroy the structural integrity of the enzymes (Cheng & Ivanov, 2012;Dong et al., 2018;Sinha & Wang, 2020), hence the systems are well adapted for other experimental investigations.

Conclusion
Of all the 83 endophytes-derived compounds evaluated in this study, asperflavin, ascomfurans C, camptothecine and corynesidone A are potent multi-target inhibitors of all the enzyme targets (AChE, BChE, BACE1 and MAO). Based on the BBB permeability, oral drugability of Lipinski rule of five and ligand-enzymes complexes stability, corynesidone A is however, the most promising candidate for further studies towards development of new neurotherapeutics.
Disclosure statement