The insight of in vitro and in silico studies on cholinesterase inhibitors from the roots of Cimicifuga dahurica (Turcz.) Maxim.

Abstract Cholinesterases (ChEs) are enzymes that break down neurotransmitters associated with cognitive function and memory. We isolated cinnamic acids (1 and 2), indolinones (3 and 4), and cycloartane triterpenoid derivatives (5–19) from the roots of Cimicifuga dahurica (Turcz.) Maxim. by chromatography. These compounds were evaluated for their inhibitory activity toward ChEs. Compound 1 was determined to have an IC50 value of 16.7 ± 1.9 μM, and to act as a competitive inhibitor of acetylcholinesterase (AChE). Compounds 3, 4 and 14 were found to be noncompetitive with IC50 values of 13.8 ± 1.5 and 6.5 ± 2.5 μM, and competitive with an IC50 value of 22.6 ± 0.4 μM, respectively, against butyrylcholinesterase (BuChE). Our molecular simulation suggested each key amino acid, Tyr337 of AChE and Asn228 of BuChE, which were corresponded with potential inhibitors 1, and 3 and 4, respectively. Compounds 1 and 4 were revealed to be promising compounds for inhibition of AChEs and BuChEs, respectively.


Introduction
Cholinesterases (ChEs), which are enzymes that hydrolyse choline esters, are classified as acetylcholinesterase (EC 3.1.1.7, AChE) and butyrylcholinesterase (EC 3.1.1.8, BuChE) 1 . AChE is responsible for the conversion of acetylcholine (ACh) into choline and acetic acid in cholinergic synapses. AChE is formed as a tetramer of $70-kDa monomeric subunits 1,2 . Its 3D structure was revealed by examining the enzyme of electric eels 3 . AChE has an active site with a-helix and b-sheet structures and a catalytic triad of serine, histidine, and glutamic acid 1,3 . BuChE, an enzyme that breaks down artificial butyrylcholine, is known to hydrolyse ACh and other ester derivatives in the body 4,5 . BuChE, which is a tetrameric serine esterase consisting of monomers of $90-kDa molecular mass, showed over 65% structural similarity to AChE 4,6 . ACh is a neurotransmitter that is produced from the acetylation reaction of choline and acetyl-CoA by choline acetyltransferase, and is distributed in the central and peripheral nervous systems 7 . ACh plays a key role in nerve-nerve communication by binding to ACh receptors 8 . This molecule is associated with maintenance of cognitive function and memory 5,8 . In particular, Alzheimer's disease (AD) patients are characterised by a decline in ACh levels 8 . Two ChEs have been regarded as target enzymes for treatment of AD 1,5,8 .
Cimicifuga dahurica (Turcz.) Maxim., in the family Ranunculaceae, is commonly called "shengma" and is distributed throughout Korea, Japan, China and Russia 9 . In China, the rhizomes of C. dahurica have been used as a traditional medicine to treat headaches and toothaches 10 . Phytochemical studies of this plant indicated the presence of cycloartane triterpenoids and cinnamic acid derivatives 9 . These compounds exhibit neuroprotective activity and enhance cell viability by eliminating H 2 O 2 in PC12 cells 9,11 . Cycloartane triterpenoids have anti-tumour activities including induction of apoptosis and G 2 /M cell cycle arrest in solid tumours, blood tumours and drug-resistant tumours 10 .
These findings led us to search for products that block the catalytic reaction of ChE. We isolated compounds 1-19 from the roots of C. dahurica using open column chromatography (CC). These compounds were tested for interactions with both AChE and BuChE in vitro. Through molecular simulation, the inhibitor-ChE complex structure was predicted visually using the Autodock 4.2 programme. The complex that was constructed considered the interaction between the inhibitor and ChE in terms of molecular dynamics (MDs).

Plant material
The roots (3.5-years old) of C. dahurica were purchased from a herbal company, Naemome Dah, Ulsan, Korea, in February 2016. This sample was identified by Prof. Y.H. Kim. A voucher specimen (CNU-16003) representing this collection has been deposited at the Herbarium of the College of Pharmacy, Chungnam National University, Daejeon, Korea.

Extraction and isolation
The roots of C. dahurica (2.5 kg) were extracted three times with 5.0 L of 95% ethanol at 40 C. Concentrated ethanol extract (65.3 g) was suspended in distilled water and progressively fractioned with n-hexane (9.6 g), dichloromethane (15.2 g) and water (40.5 g) fractions.

ChE assay
AChE and BuChE inhibition assays were performed as described by Othman et al. 12 with some modifications. Briefly, each 130 lL of AChE ($0.05 U/mL) and BuChE ($0.05 U/mL) in 50 mM phosphate buffer (pH 7.4) was added to 96-well plates containing 20 lL of MeOH or sample dissolved in MeOH. 25 lL acetylthiocholine iodide (5 mM) or butyrylthiocholine iodide (5 mM), and 25 lL DTNB (1 mM) were added into the mixture in order. After initiating ChE reaction at 37 C, the products were scanned at 475 nm UV-Vis photometer for 20 min. The inhibition activity was calculated using the following equation: Inhibitory activity ð%Þ ¼ ½ðDcontrolÀDsampleÞ=Dcontrol Â 100: Where control and sample were the intensity of control and inhibitor after 20 min, respectively.
The ChE inhibitory activity of each sample was expressed as IC 50 (mM required to inhibit the hydrolysis of the ChE substrates by 50%) determined from the log-dose inhibition curve.

Molecular docking of inhibitor with ChE
Molecular docking was performed as previously described using the Autodock 4.2 programme (La Jolla, CA, USA) 12 . Single bond of ligand was flexibly assigned by using torsion tree of Autodocktools. Each pdb files of AChE (pdb ID: 1C2B) and BuChE (pdb ID: 4BDS) were downloaded from RCSB protein data bank. Achieved protein was added in hydrogens, and then this was assigned with compute gasteiger charges. For the docking, the grid containing activity site or all protein was set. Ligand was docked into that with default values of genetic algorithm parameters (number of GA runs: 50, maximun number of evals: 25,000,000). The result was presented with Ligplot (Cambridge, UK) and Chimaera (San Francisco, CA, USA).

Molecular simulation of inhibitor with ChE
MDs were performed to simulate the complex of ligand with protein by the Gromacs version 4.6.5 package. Itp and gro files of ligand were built at Prodrg server. Gro and topology files of ChE were generated by pdbgmx utility. These were edited to add ligand files. The ligand with protein was dissolved in water molecules of a cubic box with a size of 12 Â 12 Â 12 containing six sodium ions (1.0 Å distance). Moreover, then this complex was minimised until it reach the maximal force of 10 kJ/mol. Each NVT and NPT was simulated at 300 K temperature and 1 bar pressure for 100 ps in the order, respectively. Lastly, equilibrated complex was subjected to MD simulation for 10,000 ps.

Statistical analysis
Statistical significance was determined using a one-way analysis of variance and Students t-test (Systat Inc., Evanston, IL, USA). A p value <0.01 was considered significant. All results are presented as the mean ± SEM.

Isolation and identification
An ethanol extract of the roots of C. dahurica was progressively partitioned into n-hexane, dichloromethane, and water fractions. The dichloromethane and water fractions were subjected to various CC methods to obtain compounds 1-13 and 14-19, respectively. These compounds were investigated based on spectroscopic data and comparison with previous reports. The nineteen extracted compounds were identified as cimiciphenone (1)

ChE assay
To screen for the ability of the isolated compounds 1-19 to block catalytic reaction of ChE, they were analysed in vitro at 100 lM concentration using a UV-spectrophotometer. As shown in Figure 2(A) and Table 1, compounds 1-4 and 6-8, and compounds 2-6, 9 and 14-18 exhibited over 50% inhibitory activity against AChE and BuChE, respectively. To calculate their IC 50 values, these compounds were subjected to an enzyme assay at a variety of concentrations. They caused decreases in activities of the two ChEs, with gradual or sharp slopes in activity curves in a dose-dependent manner ( Figure  2(B,C), Table 1). These results showed that compounds 1-4 and 6 -8 had IC 50 values ranging from 16.7 ± 1.9 to 95.8 ± 5.1 lM against AChE. In particular, compound 1 had an IC 50 value of 16.7 ± 1.9 lM. Compounds 2-6, 9 and 14-18 were revealed to have IC 50 values ranging from 6.5 ± 2.5 to 90.9 ± 6.0 lM against BuChE.

Enzyme kinetics on AChE and BuChE
As indicated in Figure 2(D), compound 1 was competitive inhibitor which observed to have same V max value, and different K m values at 6.2-50 lM concentration on AChE. Compounds 3 and 4 were confirmed as noncompetitive mode due to various V max values and a K m value according to respective concentrations on BuChE (Figure 2(E,F)). Whereas, compound 14 was revealed to take the binding into activity site by competing with substrate ( Figure  2(G)). Additionally, these results were calculated with K i values of the potential inhibitors using secondary replot. Compound 1 was calculated to be 16.2 ± 0.9 lM on AChE. Compounds 3, 4 and 14   were solved to be 4.9 ± 2.1, 3.5 ± 1.5 and 10.7 ± 1.3 lM on BuChE (Figure 2(H), Table 2).

Molecular docking of inhibitors with AChE and BuChE
These findings suggest that compounds 1, 3, 4 and 14 may bind with either AChE or BuChE. An inhibitor (1) was found to dock into the active site of AChE, thus acting as a competitive inhibitor of this enzyme. This inhibitor was fitted into the binding site in a stable position with an Autodock score of -9.42 kcal/mol. Compound 1 formed five hydrogen bonds (Ser203: 2.95 Å; Phe295: 2.86 Å; Phe338: 2.73 Å; His447: 2.59 Å and 2.67 Å) with four amino acids in AChE and had a hydrophobic interaction with amino acids surrounding the active site ( Figure 3(A,B)). Compounds 3, 4 and 14 exhibited molecular docking with BuChE. The noncompetitive inhibitors (3 and 4) were simulated in a blind docking test to search for possible binding with BuChE. The competitive inhibitor (14) docked into BuChE in the method described earlier. As shown in Figure 3(C-F), compounds 3 and 4 were confirmed to have hydrophobic interactions with amino acids, but not to form hydrogen bonds. The predicted binding site was proposed as the location where the inhibitor clustered with a low Autodock score. In particular, compound 4 was stably placed in the active site, with the top five positions scoring from -7.25 to 7.26 kcal/mol (Supplementary Figure S1). Enzyme kinetic results showed that compound 4 preferentially bound to the allosteric site. Therefore, the catalytic site was excluded as a binding site for this compound (4). Our results predicted the binding site as that with the lowest Autodock score for the next cluster, similar to the results for compound 3. In addition, compound 14 exhibited hydrophobic interactions with seventeen residues and formed six hydrogen bonds (Asp70: 2.97Å; Glu119: 2.71 Å; Glu276: 2.71 Å; Asn289: 3.12 Å; Trp430: 2.74 Å; Tyr440: 2.60 Å) with six amino acids in the active site of BuChE (Figure 3(G,H)).

MDs of inhibitors with AChE and BuChE
We performed MD simulations to study the stability of the inhibitor-ChE complex in solution at 300 K under 1 bar of pressure. The complexes of AChE with compound 1, and BuChE with compounds 3 and 4 were simulated stably with potential energies of about -2.35 Â 10 6 kJ/mol (Figure 4(A)). As indicated in Figure  4(

Conclusions
AD is a neurodegenerative disease caused by destruction of neurons in the central nervous system 20 . Cycloartane triterpenoids and cinnamic acid derivatives from the roots of C. dahurica have been reported to have a neuroprotective effect on PC12 cells 9,11 . The cholinergic hypothesis of AD is supported by increased memory and cognition function after binding of ACh to ACh receptors in the brain 8,21 . Therefore, AChE and BuChE are considered promising target enzymes for treating AD disease due to their effect of decreasing ACh levels 4,8 .
Our study led to isolation of cinnamic acids (1 and 2), indolinones (3 and 4), and triterpenoid derivatives (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19) from the roots of C. dahurica. We analysed these compounds to evaluate their inhibitory activity on both AChE and BuChE. Compound 1 has an IC 50 value of 16.7 ± 1.9 lM against AChE, while compounds 3 and 4 were determined to have IC 50 values of 13.8 ± 1.2 and 6.5 ± 2.5 lM against BuChE, respectively. According to BuChE assay results, triterpenoid glycosides showed more potent inhibitory activities than those of their aglycones except for compound 19. Above all, indolinone derivatives (3 and 4) were highly potential inhibitors compared to the others. In reported studies, alkaloid derivatives, such as atherosperminine, (þ)-N-methylisococlaurine, berberine, 9amino-1,2,3,4-tetrahydro acridine, and rivastigmine, were found to be famous ChEs inhibitor 12,20,21 . Moreover, blood brain barrier (BBB) plays a role to keep neuronal cells from neurotoxic substances of outside. However, BBB transporters of glucose, phenylalanine, araginine and lactate are responsible for transporting small molecules, such as deoxyglucose, galactose, lysine, pyruvate and guanosine, into brain 22,23 . Especially, compound 4 having 199 Da alkaloid may overcome the block of BBB and potentially invade into brain. Based on the enzyme kinetic study, compound 1 was shown to block catalytic reaction by interacting with the active site of AChE. Compounds 3 and 4 were revealed to have affinity for the allosteric site of BuChE. Their binding positions were predicted for the active or allosteric sites using the Autodock 4.2 package. Moreover, MD analysis led us to propose the key amino acid involved in ligand-receptor interactions. As a result, the ketone form of the ester in cimiciphenone (1) exhibited hydrogen bonding with the aromatic hydroxyl group of Tyr337 in the active site of AChE during simulation. To develop a new cinnamic acid moiety of AChE, chemists should consider compounds that are capable of interaction with Tyr337. (E)-3-(3 0 -methyl-2 0 -butenylidene)-1-methyl-2-indolinone (3) and (E)-3-(3 0methyl-2 0 -butenylid-ene)-2-indolinone (4) participated in hydrogen bonding with Asn228 located at its predicted binding site on BuChE. It is necessary to develop a prenyl indolinone derivative as a noncompetitive inhibitor to enhance hydrogen bonding with this polar amino acid (Asn228). In our research, we identified the key amino acids, which could not be found through molecular docking, using MD analysis. Among cycloartane triterpenoids and cinnamic acid derivatives that show neuroprotective activity 9,11 , cimiciphenone (1), (E)-3-(3 0 -methyl-2 0 -butenylidene)-1-methyl-2-indolinone (3) and (E)-3-(3 0 -methyl-2 0 -butenylidene)-2-indolinone (4) showed promise as potential inhibitors of AChE and BuChE, respectively. Compound 1 was determined to be the optimal compound for development as a competitive inhibitor of AChE, while compounds 3 and 4 may provide a new skeleton for noncompetitive inhibitors of BuChE.