Novel 2-pheynlbenzofuran derivatives as selective butyrylcholinesterase inhibitors for Alzheimer’s disease

Alzheimer’s disease (AD) is a neurodegenerative disorder representing the leading cause of dementia and is affecting nearly 44 million people worldwide. AD is characterized by a progressive decline in acetylcholine levels in the cholinergic systems, which results in severe memory loss and cognitive impairments. Expression levels and activity of butyrylcholinesterase (BChE) enzyme has been noted to increase significantly in the late stages of AD, thus making it a viable drug target. A series of hydroxylated 2-phenylbenzofurans compounds were designed, synthesized and their inhibitory activities toward acetylcholinesterase (AChE) and BChE enzymes were evaluated. Two compounds (15 and 17) displayed higher inhibitory activity towards BChE with IC50 values of 6.23 μM and 3.57 μM, and a good antioxidant activity with EC50 values 14.9 μM and 16.7 μM, respectively. The same compounds further exhibited selective inhibitory activity against BChE over AChE. Computational studies were used to compare protein-binding pockets and evaluate the interaction fingerprints of the compound. Molecular simulations showed a conserved protein residue interaction network between the compounds, resulting in similar interaction energy values. Thus, combination of biochemical and computational approaches could represent rational guidelines for further structural modification of these hydroxy-benzofuran derivatives as future drugs for treatment of AD.

The inhibition results of the compounds against the two enzymes are summarized in Table 1. We noted that compound 28, with three hydroxyl substituents in phenyl-ring and hydrogen atom in position 5 (R) and 7 (R 1 ) of benzofuran scaffold did not exert any cholinesterase inhibitory activity. In general, except compounds 23 and However, with respect to eqBChE case, these compounds displayed a lower inhibitory activity against hBChE. Nevertheless, IC 50 values for the compounds obtained against hBChE enzyme is nearly 2 times lower to that obtained for the reference compound galantamine in the same assay conditions. Antioxidant activity assessment. The antioxidant property of compounds 15-28 was evaluated by ABTS + assay and the results are represented as EC 50 values in Table 3.
We used Trolox as positive control to compare the antioxidant capacity of the subjected compounds. All the compounds were found to possess an ability to quench ABTS radical and displayed a scavenging activity better   Cytotoxicity assay analysis. After obtaining encouraging results from the inhibitory assay experiments, biosafety effectiveness of the two promising compounds (15 and 17) was further evaluated. Cells were treated with different concentration of each compound (0-100 μM) for 24 h and their potential cytotoxic effect on NSC-34 cells was determined by using MTT assay 51 . Viability of the cells treated with the compounds 15 and 17 and comparison to the control cells were performed (Fig. 4). Moreover, results also indicated that compounds 15 and 17 exhibited no considerable cytotoxic effect in NSC-34 cells at the concentration in which eqBChE activity was inhibited.   Molecular modeling studies. To predict how the compounds 15 and 17 bind to hBChE and to understand the molecular origin of their high inhibitory activity and selectivity, we performed molecular docking experiments. Docking results suggested similar interaction sites (Fig. 5a,b) and similar binding energy values (~7.5 kcal/ mol), for the two compounds. The stability of the docking poses of the two compounds was investigated using MD simulations, which is a standard technique used to study the dynamical properties of biomolecules [52][53][54][55][56] . The stability of the systems during the MD simulations was evaluated by calculating the root mean square deviation (RMSD) of C-alpha atoms of protein residues (Fig. 5c) from the starting structure. The average RMSD values of protein bound compound simulations were lower than in free protein simulations, with lowest value noted for compound 17 complex simulations. Subsequently, the interaction energy between the hBChE residues and the two compounds was calculated by evaluating the non-bonded energy values comprising of Van der Waals and electrostatic energy in the two simulations. Both the complexes exhibited similar interaction energy values (Fig. 5d).  To understand the origin of this similarity, we carefully inspected the binding mode of the compounds in complex with hBChE using Ligplot 57 . The compounds (15 and 17) were stably bound to hBChE active site ( Fig. 6) encompassing the region between peripheral anionic site (PAS) and the catalytic triad site (CAS). Figure 6 depicts five overlapping hBChE residues interacting with the two compounds. In detail, these residues are located in catalytic triad (S198), oxyanion hole (G117), acyl-pocket (L286, V288) and wall of BChE active site. The hydroxyl substituents in compound 17 interact with peripheral anionic site residue (Y332), while compound 15 interacts with oxyanion hole residue (G116) and residue T120.
To examine the effects of compound 15 and 17 on the protein structural dynamics, comparative analysis of a series of snapshots of the protein coordinates from MD simulations trajectories between the complex (bound to the compounds) and free protein was done. Calculation of all inter-residue cross-correlations fluctuations (see Methods) of C-alpha atoms resulted in a matrix of cross-correlation coefficient (C ij ) elements, which are displayed in a graphical representation as a dynamical cross-correlation map, shown in Fig. 7.
As expected, we note strong fluctuations occur along the diagonal occur (between the same residue), wherein C ij is always equal to 1. A clear difference in the cross-correlations maps between the free and complex simulations was observed (Fig. 7). With respect to free protein simulations (Fig. 7a), we observed between few domains, an increase in either a positive or a negative correlation dynamics for the complex simulations (Fig. 7b,c). In detail, the regions involved in higher negative correlated dynamics included residues 40-60, 170-190 and 380-500, while residues 230-280 displayed lower negative correlated dynamics. On the other hand, residues 430-470 exhibited higher positive correlated dynamics in the compound complexes. As expected, most of these regions are in close vicinity to the hBChE active site gorge. Interestingly, only for compound 17 complex (Fig. 7b), positive correlation dynamics was noted between the domains surrounding the BChE active site gorge, i.e. residues 240-280 and 300-330, respectively.

Discussion
There is increasing clinical evidence suggesting an important role of BChE in the regulation of ACh levels and in particular in the development and progression of AD. Particularly, in progressed or late stage of AD, BChE mostly dominates hydrolysis of ACh 58 . Moreover, alongside its involvement in AD progression, an emerging role of BChE as a prognostic marker (which determines the progress of the disease) in liver and non-liver diseases, as well as in protein-energy malnutrition and obesity, has been reported 15,59 . Design and development of compounds with the ability to selectively inhibit BChE would not only improve understanding of the aetiology of AD but also assist in developing wider variety of new treatments. Therefore, the objective of our study has been to design and develop 2-phenylbenzofuran compounds that display selective BChE inhibitory activity employing biochemical, kinetics and computational techniques.
In our recent study 34 , we reported that the contemporary presence of a hydroxyl group in the para position of the 2-phenyl ring and a halogen substitution at position 7 (R 1 ) of the benzofuran scaffold resulted in a good and selective BChE inhibition, with best inhibitor displaying an IC 50 of 30 µM. Following the results of our previous findings, in this present work we decided to explore the importance of the number and position of hydroxyl groups located in the 2-phenyl ring of the benzofuran moiety. We therefore synthesized new 2-phenylbenzofurans compounds with two hydroxyl substituents (compounds 15-21) and with three hydroxyl substituents (compounds 22-28). Galantamine was used as our reference compound. The inhibitory action of the newly synthesized compounds presented in Table 1 demonstrate that, regardless the type of substituent at position 7 of benzofuran scaffold, the 2-phenylbenzofuran derivatives with two hydroxyl substituents (compounds

15-21)
in meta position of the 2-phenyl ring displayed rather high inhibitory activity toward eqBChE and very low activity against EeAChE. In particular, compounds 15 and 17 displayed eqBChE inhibitory activity 4-and 8times more effective than the reference compound, respectively. However, in the compounds with three hydroxyl substituents (instead of two) in the 2-phenyl ring (compounds 22, 24), we found lower inhibitory activity against eqBChE. This fact suggest that contemporary presence of three hydroxyl groups in the 2-phenyl ring of the compounds could decrease the inhibitory activity of the compounds against eqBChE. It has been shown previously 60 that the position and number of hydroxyl group in the ligand can influence the magnitude of hydrogen bond interactions with the protein. The BChE active site is located at the bottom of a 20 Å deep gorge that is lined mostly with hydrophobic residues. Thus, binding of an additional hydroxyl substituent (a polar group) within the gorge could result in a thermodynamic penalty of additional 4.3-5.3 kcal/mol 61 , due to energetic cost of desolvation. Hence, this could be one possible hypothesis to explain the low BChE inhibitory activity detected for the compounds with three hydroxyl groups in the 2-phenyl ring.
The two most active compounds (15,17) differ in halogen atom at position 7 of the benzofuran moiety (chlorine, bromine atoms), respectively. It is interesting to note that this little difference is reflected in the protein interaction network characterizing these compounds (Fig. 6). The chlorine atom in compound 15 interacts with the CAS residue (S198) and F398, while bromine atom in compound 17 interacts with the acyl pocket residues (L286, V288). The ChE inhibition can occur either via a competitive interaction with CAS, or a non-competitive binding with PAS, or via mixed-type mechanisms, by exerting a dual binding ChE inhibition 62 . Enzyme kinetic analysis (Figs 2 and 3) demonstrated only compound 15 as mixed-type inhibitor, while compound 17 as non-competitive inhibitor of eqBChE activity. The results from kinetic experiments are confirmed from MD simulations, which provide molecular-level insights into how ligand binding at an allosteric site can affect protein structure and, consequently, enzymatic activity. Indeed, the difference in the nature of correlated protein dynamics (Fig. 7) noted between the compound 17 and compound 15 complexes (Fig. 7b), could possibly explain their different inhibition mechanisms against BChE. In detail, only for compound 17 complex, we observed a positive correlated motion between the domains surrounding the BChE active site gorge, i.e. residues 240-280 and 300-330, respectively.
Previous clinical studies evidence that oxidative stress is a crucial factor in AD and plays an important role in inducing and activating multiple cell signalling pathways, contributing to the development of AD 63,64 . Indeed, development of new avenues to reduce oxidative damages can provide therapeutic efficacy in the treatment of AD 65 . We therefore investigated the antioxidant properties of the new synthesized compounds. Comparing the results with the antioxidant property of benzofurans derivatives analyzed in our previous study 34 , compounds 15 and 17 showed a higher antioxidant activity (Table 3). Thus, substitution and positioning the groups within the 2-phenyl ring of the compounds, led to an improvement in terms of both BChE inhibitory activity and antioxidant property.

Conclusions
In this study, a series of hydroxylated 2-phenylbenzofurans compounds were designed, synthesized and their selective inhibitory activity BChE was evaluated. Combining biochemical analysis and computational approaches, we identified two potent BChE inhibitors as compound 17 (IC 50   than the reference compound. In conclusion, gathering the information obtained in this study, compounds 15 and 17 could be considered as promising candidates for the design and development of drugs against AD.

Chemistry. Starting materials and reagents were obtained from commercial suppliers (Sigma-Aldrich) and
were used without further purification. Melting points (mp) are uncorrected and were determined with a Reichert Kofler thermopan or in capillary tubes in a Büchi 510 apparatus. 1 H NMR and 13 C NMR spectra were recorded with a Varian INOVA 500 spectrometer using [D 6 ]DMSO or CDCl 3 as solvent. Chemical shifts (δ) are expressed in parts per million (ppm) using TMS as an internal standard. Coupling constants J are expressed in hertz (Hz). Spin multiplicities are given as s (singlet), d (doublet), dd (doublet of doublets), and m (multiplet). Elemental analyses were performed by using a Perkin Elmer 240B microanalyzer and are within 0.4% of calculated values in all cases. The analytical results indicate 98% purity for all compounds. Flash chromatography (FC) was performed on silica gel (Merck 60, 230-400 mesh); analytical TLC was performed on pre-coated silica gel plates (Merck 60 F254). Organic solutions were dried over anhydrous Na 2 SO 4 . Concentration and evaporation of the solvent after reaction or extraction was carried out on a rotary evaporator (Büchi Rotavapor) operating under reduced pressure.

Preparation of Hydroxylated 2-phenylbenzofurans.
A solution of the corresponding methoxy-2phenylbenzofuran (0.11 g, 0.50 mmol) in acetic acid (5.0 mL) and acetic anhydride (5.0 mL), at 0 °C, was prepared. Hydriodic acid 57% (10.0 mL) was added drop-wise. The mixture was stirred under reflux temperature for 3 h. The solvent was evaporated under vacuum and the dry residue was purified by FC (dichloromethane/methanol 9.8:0.2) to give the desired compound 15-28 47,49,50 . Cholinesterase assay. The enzymes and reagents for biochemical assays were obtained from Sigma-Aldrich.
Kinetic assays of cholinesterase activity were performed using Ellman's method and analyzed as previously described 34 . Briefly, in the microplate assay the reaction mixture contained phosphate buffer (0.1 M, pH 8.0), AChE or BChE solution (0.3 or 0.15 U/mL respectively), 5,5′-dithiobis-(2-nitrobenzoic) acid (DTNB; 1.5 mM), and inhibitor dissolved in 1% DMSO at the desired concentrations or DMSO alone (control) in a final volume of 0.2 mL. Finally, acetylthiocholine iodide (ATCI) or S-butyrylthiocholine iodide (BTCI) (1.5 mM) as the substrate was added to the reaction mixture and the absorbance immediately monitored at 405 nm. The activity of the enzymes was performed at 25 °C.
Acetylcholinesterase was from Electrophorus electricus (EeAChE), while butylcholinesterase was from equine serum (eqBChE) or human serum (hBChE). Each inhibitor was evaluated at six concentrations (ranging from 0.5 to 100 µM). Galantamine was used as the standard cholinesterase inhibitor.
The inhibition potency was expressed in IC 50 values, which represent the inhibitor concentration giving 50% inhibition of cholinesterase activity. IC 50 values were calculated by the interpolation of dose-response curves using GraphPad Prism 6 (Graphpad Software, San Diego, California, USA). IC 50 values displayed represent the mean ± standard deviation for three independent assays.
Kinetic characterization was performed by constructing Lineweaver-Burk plots by plotting 1/V vs 1/[S] in the presence of different concentrations of inhibitor and substrate. Kinetics constants were determined by the replots of the Slopes (K M /V max ) or 1/V max versus the inhibitor concentration.
Antioxidant activity. Total free radical-scavenging capacity of the compounds was determined by ABTS .+ [2,2′-azinobis-(3-ethylbenzothiazoline-6-sulfonic acid)] method using Trolox as antioxidant standard, as previously reported 66 . Briefly, the free radical ABTS .+ was produced by reacting 7 mM ABTS with 2.45 mM potassium persulfate in aqueous solution and kept in the dark for 24 h at room temperature before use. After appropriate dilution, each compound (10 μL) was added to 1 ml of ABTS .+ solution and the absorbance at 734 nm was recorded after 1 min incubation. Results were expressed as EC 50 values (μM), the concentration of sample necessary to give a 50% reduction in the original absorbance.
Cell viability. Mouse motor neuron like cell line (NSC-34) was purchased from the American Type Culture Collection (ATCC, Manassas, Virginia, USA). The cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) containing 10% fetal bovine serum (FBS; Gibco, NY, USA), and 1% penicillin/streptomycin at 37 °C in a humidified atmosphere with 5% CO 2 . Cell viability was detected by the colorimetric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay 51 . This is a colorimetric assay for measuring the activity of mitochondrial enzymes in living cells that convert MTT into purple formazan crystals. Briefly, cells were seeded in a 96-well plate (10 4 cells/well) and incubated with samples at concentration ranging from 10 to 100 μM for 48 h. As DMSO was used as solvent for compounds, all activities were performed also in the presence of DMSO alone, as solvent control. After incubation time, cells were labelled with MTT solution for 3 h at 37 °C. The resulting violet formazan precipitates were dissolved in isopropanol and the absorbance of each well was determined at 590 nm using a microplate reader with a 630 nm reference.

Molecular Modeling.
High-resolution three-dimensional protein structure of hBChE was obtained from protein data bank (PDB id: 4TPK). For the compounds (15 and 17), the three-dimensional coordinates were SCIentIfIC RepoRts | (2018) 8:4424 | DOI:10.1038/s41598-018-22747-2 generated using Open Babel software 67 . The geometry of the compounds were optimized using the Hartree-Fock basis set 6-31 G* within Gaussian03 software package 68 . The charges and the force field parameters of the compounds were evaluated following the standard protocol within AMBER software tools 69,70 .
Molecular docking of the compounds into hBChE protein was performed using SwissDock web server, which is based on the docking software EADock DSS 71 . The docking poses of the compounds were accurately chosen with a blind docking procedure that considers the entire protein surface as a potential target. Using this procedure, a large number of ligand binding modes (~15000) were generated, with the simultaneous rough interaction energy estimation. The binding modes possessing favorable energies were then ranked and classified into different clusters, this time based on the full fitness scoring function. The most consistent and favorable conformation chosen from 10 independent docking runs for each compound was further considered for MD simulations.
The hBChE-compound complexes were built using leap module of Amber11. Each complex was inserted separately in an explicit water-box with a minimum distance of 1.8 nm between the solute and box boundary. Further details about the simulation box size and the total number of atoms for each complex are provided in Supplementary Table S1. We used amber force-field parameters 72 for hBChE protein and TIP3P 73 parameters for water molecules. Energy minimization, followed by heating of the complexes to temperature 300 K, was done with positional restraints on C-alpha atoms. The positional restraints were gradually removed during the simulation time and an equilibration run of 10 ns was performed. The time step used in MD simulation was of 2 fs using SHAKE algorithm. Simulations were performed in NPT ensemble using periodic boundary conditions. All-atom MD simulations of free protein and protein-compound complexes were performed for a simulation time of 100 ns employing NAMD 74 software package.
The stability of systems was evaluated by calculating the RMSD values for the C-alpha atoms of residues during MD simulations, using VMD 75 . The interaction energy between the compound and protein residues was calculated by evaluating the non-bonded energy values comprising of Van der Waals and electrostatic energy, using the energy plugin of NAMD software. A cut-off distance of 12 Å was used for non-bonded interactions and for the electrostatic interaction we also adopted the particle mesh Ewald 76 scheme. The dynamic cross-correlation 77 coefficients for C-alpha atoms was calculated on 1000 snapshots extracted from 100 ns MD trajectories using Prody 78 software. The matrix of all inter-atomic cross-correlations of atomic fluctuations C ij where i and j are C-alpha atoms, can be represented as a dynamical cross-correlation map. If the fluctuations of two C-alpha atoms are completely correlated then C ij = 1 (red), if anticorrelated then C ij = −1 (blue), and if C ij = 0 (white) then the fluctuations of i and j are not correlated.