Acetylcholinesterase and butyrylcholinesterase inhibitory activities of khellactone coumarin derivatives isolated from Peucedanum japonicum Thurnberg

Cholinesterase (ChE) and monoamine oxidase (MAO) inhibitors have been attracted as candidate treatments for Alzheimer's disease (AD). Fifteen khellactone-type coumarins from the roots of Peucedanum japonicum Thunberg were tested for acetylcholinesterase (AChE), butyrylcholinesterase (BChE), and MAO inhibitory activities. Compound 3′-angeloyl-4′-(2-methylbutyryl)khellactone (PJ13) most potently inhibited AChE (IC50 = 9.28 µM), followed by 3′-isovaleryl-4′-(2-methylbutyroyl)khellactone (PJ15) (IC50 = 10.0 μM). Compound senecioyl-4′-angeloyl-khellactone (PJ5) most potently inhibited BChE (IC50 = 7.22 μM) and had the highest selectivity index (> 5.54), followed by 3′-senecioyl-4′-(2-methylbutyryl)khellactone (PJ10) and 3′,4′-disenecioylkhellactone (PJ4) (IC50 = 10.2 and 10.7 μM, respectively). Compounds PJ13, PJ15, and PJ5 showed reversible and mixed-types of inhibition with Ki values of 5.98, 10.4 (for AChE), and 4.16 µM (for BChE), respectively. However, all 15 compounds weakly inhibited MAO-A and MAO-B. Molecular docking simulation revealed that PJ13 had a higher binding affinity (− 9.3 kcal/mol) with AChE than PJ15 (− 7.8 kcal/mol) or PJ5 (− 5.4 kcal/mol), due to the formation of a hydrogen bond with Tyr121 (distance: 2.52 Å). On the other hand, the binding affinity of PJ5 (− 10.0 kcal/mol) with BChE was higher than for PJ13 (− 7.7 kcal/mol) or PJ15 (− 8.1 kcal/mol), due to the formation of a hydrogen bond with Ser198 (distance: 2.05 Å). These results suggest that PJ13 and PJ5 are potential reversible selective inhibitors of AChE and BChE, respectively, for the treatment of AD.


Materials and methods
Compounds. Fifteen khellactone-type compounds were isolated from P. japonicum Thunberg (voucher specimen: PBC-484), and the structures were determined, as described previously 39 . Briefly, the dried roots of P. japonicum (5.0 kg) were extracted with 80% ethanol (EtOH) at room temperature three times to obtain 1.62 kg of solid extract. The 80% EtOH extract was further partitioned between n-hexane (114.2 g) and H 2 O (1.50 kg), and the n-hexane extract so obtained was subjected to preparative reverse phase chromatography (Xbridge Prep C 18 [1][2][3][4][5][6][7][8] were collected and concentrated on a rotary evaporator under reduced pressure. Purification was conducted by recycling preparative HPLC. The yield of the khellactone-type coumarins obtained was ~ 1.5% from 80% EtOH extract determined by using ultra-performance liquid chromatography (UPLC) charged with photodiode array (PDA). Chemical structures of the compounds were identified by 1 H NMR, 13  Enzyme assays. AChE assays were performed as described by Ellman et al. 41 with slight modifications 42 . In brief, assays were performed using ~ 0.2 U/mL of AChE in the presence of 0.5 mM DTNB and 0.5 mM ACTI in 0.5 mL reaction mixtures, and continuously monitored for 10 min at 412 nm. DTNB was used for color development, caused by reaction between it and thiocholine (a product of AChE). For inhibitory assays, compounds were preincubated with AChE for 15 min prior to ATCI and DTNB addition. BChE activity was assayed using the same method, but BTCI was used instead of ATCI. MAO-A activity was continuously assayed using kynuramine (a substrate) at 316 nm for 20 min, and MAO-B activity was assayed using benzylamine at 250 nm for 30 min, as described previously 40    Analysis of pharmacokinetic properties using in silico method. Drug-like properties of the lead compounds of PJ5, PJ13, and PJ15 were analyzed using a web tool of SwissADME at http://www.swiss adme. ch/ 48 . www.nature.com/scientificreports/

Results
Analysis of inhibitory activities. The structures and purities of the 15 compounds isolated from the P.
japonicum Thunberg, were determined by 1D and 2D NMR spectra, UPLC-QTOF-MS analysis, and electronic circular dichroism spectra 39 . All were tested for AChE and BChE inhibitory activities at a concentration of 10 µM. PJ13 and PJ15 resulted in AChE residual activity of < 50% (Table 1). PJ13 most potently inhibited AChE with an IC 50 value of 9.28 µM, followed by PJ15 and PJ7 (IC 50 = 10.0 and 17.9 µM, respectively). The other 12 compounds had IC 50 values of ≥ 20 µM. In addition, four compounds resulted in BChE residual activity of < 50% (Table 1). PJ5 most potently inhibited BChE with an IC 50 value of 7.22 µM, followed by PJ10 PJ4, and PJ9 (IC 50 = 10.16, 10.66, 12.5 µM, respectively) ( Table 1). The other 11 compounds had IC 50 values of ≥ 40 µM. PJ5 had the highest selectivity index of > 5.54. To examine the multi-targeting abilities of the compounds, we evaluated their inhibitory effects on MAO-A or MAO-B, which are auxiliary targets in AD. However, all compounds only weakly inhibited MAO-A or MAO-B with residual activities of > 63.1% at 10 µM ( Table 1).

Reversibilities of AChE and BChE inhibitions.
Inhibitory assays were carried out after preincubating AChE or BChE with inhibitors for 15 min. The reversibilities of AChE inhibitions by PJ13 and PJ15 were investigated using a dialysis-based method. Inhibitions of AChE by PJ13 and PJ15 recovered from 34.7% (A U ) to 72.3% (A D ) and from 32.8% to 68.7%, respectively, which were similar to those shown by tacrine (from 14.7% to 73.6%), a reversible AChE inhibitor ( Fig. 2A). In addition, inhibition of BChE by PJ5 recovered from 41.2% (A U ) to 86.8% (A D ), which was similar to that of tacrine (from 29.9% to 100%), also a reversible BChE inhibitor (Fig. 2B). These results indicate that PJ13 and PJ15 are reversible inhibitors of AChE and PJ5 is a reversible inhibitor of BChE.
Analysis of inhibitory patterns. Modes of AChE inhibitions by PJ13 and PJ15 were investigated using Lineweaver-Burk plots. Plots of AChE inhibition by PJ13 were linear and lines intersected at a point, but not at the x-or y-axis (Fig. 3A). Secondary plots of the slopes of Lineweaver-Burk plots against inhibitor concentrations showed that the K i value of PJ13 for AChE inhibition was 5.99 ± 0.21 µM (Fig. 3B). Plots of AChE inhibitions by PJ15 were also linear and did not intersect at the x-or y-axis (Fig. 3C), and the K i value of PJ15 for the AChE inhibition was 10.41 ± 0.67 μM (Fig. 3D). These results show PJ13 and PJ15 acted as mixed-type inhibitors of AChE. In addition, plots of BChE inhibition by PJ5 were linear and intersected near the y-axis (Fig. 3E). Secondary plots showed the K i value of PJ5 for BChE inhibition was 4.16 ± 0.72 µM (Fig. 3F), showing PJ5 acted as a mixed-type BChE inhibitor.  The results of the docking simulation for AChE showed that PJ13 interacted by forming a hydrogen bond with Tyr121 (distance: 2.52 Å). However, no hydrogen bond interaction was predicted for PJ5 and PJ15 (Fig. 4A-C). Docking simulation of PJ5 with BChE implied that a hydrogen bonding interaction was established with Ser198 (distance: 2.05 Å) of BChE, whereas no hydrogen bond was proposed for PJ13 and PJ15 (Fig. 4D-F). The binding affinity of PJ13 (− 9.3 kcal/mol) for AChE was higher than that of PJ15 (− 7.8 kcal/mol) or PJ5 (− 5.4 kcal/mol) ( Table 2). In addition, PJ5 had higher binding affinity for BChE (− 10.0 kcal/mol) than PJ13 (− 7.7 kcal/mol) or PJ15 (− 8.1 kcal/mol). The binding affinities of PJ5, PJ13, and PJ15 with MAO-A or MAO-B were predicted to be weaker than those with AChE or BChE (Table 2). Docking simulations were provided in Supplementary Figure S17 (A-F). The binding score (− 4.8 kcal/mol) of PJ13 for MAO-B was relatively higher than those of PJ5 and PJ15 in accordance with the residual activities at 10 µM. When the crystal structure of AChE complexed with donepezil (PDB ID: 6O4W) and the binding pockets for BChE, MAO-A, and MAO-B defined with donepezil were used for docking simulations, the binding scores of PJ compounds were similar to the values obtained with their complexed ligands (Tables 2 and Supplementary  Table S3). From docking simulations with the AChE/donepezil complex (PDB ID: 6O4W), it was predicted that PJ13 and PJ15 formed one hydrogen bond with Tyr124 (distances = 2.602 and 2.994 Å, respectively), but PJ5 did not form the bond. On the contrary, PJ5 could form a hydrogen bond with Thr120 of BChE (distance = 3.354 Å), but PJ13 and PJ15 did not form (Supplementary Fig. S18).   www.nature.com/scientificreports/ Pharmacokinetic properties using in silico method. From the SwissADME analysis, it was predicted that the lead compounds of PJ5, PJ13, and PJ15 had high gastrointestinal adsorption abilities and cytochrome P450 inhibitory activities for 2C19, 2C9, and 3A4, however, they did not have blood-brain barrier (BBB) permeabilities (Table 3). Regarding BChE inhibition, PJ5 (IC 50 = 7.22 µM) was the most potent inhibitor, followed by PJ10 and PJ4 (IC 50 = 10.16 and 10.66 µM, respectively). The IC 50 value of PJ5 in this study was lower than those of broussonin A (7.50 µM) from Anemarrhena asphodeloidesa 42 , isoacteoside (29.7 µM) from H. procumbens 52 , corenone B (10.9 μg/mL, i.e., 49.5 µM) from Niphogeton dissecta 60 , and kaempferol (62.5 µM) from Cleistocalyx operculatus 61 , but higher than that of 4′-hydroxy Pd-C-III (5.78 µM) from A. decursiva 58 . Compared to other coumarins, the IC 50 value of PJ5 for BChE inhibition was lower than those of hyuganin C (38.86 µM), from Mutellina purpurea 62 , a coumarin pteryxin (12.96 μg/mL, i.e., 33.5 µM) from M. purpurea 63 , the esculetin (9.29 µM) and the daphnetin (8.66 µM) 56 , and it might be concluded that PJ5 is the most potent BChE inhibitor in natural coumarins reported.

Before dialysis A fter dialysis
These results show that PJ5 is potent and selective inhibitor of BChE, and that PJ13 and PJ15 are selective inhibitors of AChE. It might be suggested that combination of compounds effectively inhibit ChE. The possibility of dual inhibition of AChE and MAO enzymes was investigated for dual-or multi-targeting therapeutic purposes in AD 15,[17][18][19] . However, in the present study, no tested khellactone coumarin showed dual inhibitory activity.
AChE or BChE inhibitors have been reported to exhibit competitive, noncompetitive, and mixed-type inhibitory patterns 42,58 . In the present study, potent inhibitions of AChE by PJ13 and PJ15 and of BChE by PJ5 were found to be reversible and to exhibit mixed-type inhibition, with K i values of 5.98, 10.4, and 4.16 µM, respectively. These results suggest that PJ13, PJ15, and PJ5 bind to the allosteric site or the substrate-binding site of AChE.  www.nature.com/scientificreports/ Docking simulation analysis with AChE revealed that the PJ13 interacted with the phenolic hydroxyl group of Tyr121 to form a hydrogen bond, while no hydrogen-bond was predicted for PJ5 and PJ15. In addition, the oxygen of the carboxyl group of PJ5 formed a hydrogen bond with Ser198 of BChE, whereas no hydrogen bonding was suggested for PJ13 and PJ15. These results imply that the existence of the hydrogen bond in the complex has major effects on binding energies. Furthermore, the results concur with the K i values and binding affinities of AChE or BChE for PJ5, PJ13, or PJ15.
To explain the reason PJ15 inhibits AChE more selectively than PJ5, Van der Waals (VDW) distances and interactions were examined at C16, C17, C18, and C19 (for PJ15) or C21 (for PJ5) atoms in the docked ligands, according to the difference between PJ15 and PJ5, i.e., the 2-methyl-butane and the 2-methyl-butene group, respectively (Figs. 1 and Supplementary Fig. S16). It was predicted that thirteen and five VDW interactions were formed with PJ15 and PJ5, respectively, within a distance of 4 Å (Supplementary Table S1and S2). The VDW interactions of PJ15 could inhibit AChE more selectively than JP5.
In molecular dynamics analysis, average root mean square deviation (RMSD) values of PJ5, PJ13, and PJ15 for AChE were estimated to be 0.767, 0.684, and 0.752 Å, respectively, and those for BChE were 0.738, 0.823, 0.757 Å, respectively (Supplementary Figure S19). The results supported well the experimental data and the docking simulations in this study.
In a previous study, it was observed that PJ5, PJ13, and PJ15 were non-toxic up to 10 µg/µL (i.e., ~ 25 mM) and exhibited potent for anti-inflammatory effects at 10 µg/µL in previous study 39 , which suggests PJ5, PJ13, and PJ15 be considered candidates for the treatment of AD as ChE inhibitors with anti-inflammatory activities.

Conclusion
Among the fifteen khellactone coumarin compounds isolated from P. japonicum Thunberg, PJ5 and PJ13 were found to potently and effectively inhibited BChE and AChE, respectively. Furthermore, these inhibitors were reversible and caused by mixed inhibition. Molecular docking simulations showed that PJ13 had the highest binding affinity for AChE at − 9.3 kcal/mol, and that PJ5 had the highest binding affinity for BChE at − 10.0 kcal/ mol. These results supported the notion that PJ13 and PJ5 should be considered novel, potent, and selective inhibitors of AChE and BChE, respectively. In addition, our findings suggest that PJ5, PJ13, and PJ15 are nontoxic, reversible AChE and BChE inhibitors and candidates for the treatment of AD.