New semicarbazones as gorge-spanning ligands of acetylcholinesterase and potential new drugs against Alzheimer’s disease: Synthesis, molecular modeling, NMR, and biological evaluation

Two new compounds (E)-2-(5,7-dibromo-3,3-dimethyl-3,4-dihydroacridin-1(2H)-ylidene)hydrazinecarbothiomide (3) and (E)-2-(5,7-dibromo-3,3-dimethyl-3,4-dhihydroacridin-1(2H)-ylidene)hydrazinecarboxamide (4) were synthesized and evaluated for their anticholinesterase activities. In vitro tests performed by NMR and Ellman’s tests, pointed to a mixed kinetic mechanism for the inhibition of acetylcholinesterase (AChE). This result was corroborated through further docking and molecular dynamics studies, suggesting that the new compounds can work as gorge-spanning ligands by interacting with two different binding sites inside AChE. Also, in silico toxicity evaluation suggested that these new compounds can be less toxic than tacrine.


Introduction
The neurodegenerative diseases are characterized by progressive and irreversible loss of neurons from specific regions of the brain, affecting mainly the subcortical areas (nuclei of the base) and cerebral cortex, resulting in abnormality in the control of the voluntary movement, impairment of memory and cognitive ability. The progression of these debilitating and incurable conditions gives rise to dementia.
The estimated global number of people with some dementia is currently expected to be around 46 million and should increase to more than 130 million by 2050 (Eghtedari et al., 2017;Kuca et al., 2016;Prince et al., 2015). Alzheimer's disease (AD) is, currently, the most common cause of dementia and death among aged people (Katzman, 1976;Wilson et al., 2012). This is a multi-factorial disease with a complex etiology, requiring efforts from different areas to elucidate its molecular basis, which are still unclear today. AD primarily affects the cholinergic system and its pathological characteristics include extracellular deposition of β-amyloid plaques, intracellular formation of neurofibrillary tangles (NFTs), hyperphosphorylation and aggregation of the tau protein, neuroinflammation, and neurodegeneration (Singh, Srivastav, Yadav, Srikrishna, & Perry, 2015).
AD affects the cholinergic system due to the low production of acetylcholine (ACh), the main neurotransmitter related to the mechanisms of memory and learning. Solving this problem and, thereby slowing the progress of the disease (Auld, Kornecook, Bastianetto, & Quirion, 2002;Konrath et al., 2012;Mufson, Counts, Perez, & Ginsberg, 2008;Paul, Jeon, Bizon, & Han, 2015) was the goal of the first drugs approved for AD treatment, known as cholinesterase inhibitors (ChEIs).
Today among the five drugs approved by the Food and Drugs Administration (FDA) for the treatment of AD, three (donepezil, galantamine, and rivastigmine) are ChEIs (Choudhary, 2015). One (memantine) is a noncompetitive N-methyl-D-aspartate receptor antagonist that regulates the activity of a different chemical messenger in the brain that is also important for learning/memory and, the last one, namzaric, is a combination of donepezil and memantine (Choudhary, 2015). Another ChEI, tacrine, was the first drug approved by FDA, to be used on a large scale to treat AD as potent inhibitor of acetylcholinesterase (AChE). However, it requires four daily applications and causes hepatic alterations in 30-40% of patients. For those reasons, tacrine fell into disuse after the development of the ChEIs mentioned above (Crismon, 1994).
Once the drugs available today against AD only provide a symptomatic, palliative pharmacological effect that becomes moderate and wears off with continuous use (Bartus & Dean, 1982;Coyle, Price, & DeLong, 1983;Whitehouse et al., 1982) there is a general perception that drugs acting only on a specific target ('one molecule one target' paradigm) are no more appropriate for AD therapy. This is comprehensive considering the multi-factorial and complex nature of this disease. Therefore, approaches based on the development of multitarget direct ligands (MTDL) involved in the pathology of AD, like β-amyloid plaque and tau protein, have recently emerged (Agis-Torres, Sollhuber, Fernandez, & Sanchez-Montero, 2014;Carmo Carreiras, Mendes, Jesus Perry, Paula Francisco, & Marco-Contelles, 2013;Rosini et al., 2008). However, although MTLD are interesting options to restrain the neuronal degeneration, it is not possible yet to give up on ChEIs, since clinical symptoms of AD are detected only when the disease is already established.
Studies suggested that the formation of β-amyloid plaques in the brain is a key neurodegenerative event in AD (Dickerson et al., 2005;Murphy & LeVine, 2010;O'Brien & Wong, 2011;Zhao & Lukiw, 2015). Small molecules capable of binding to the peripheral anionic site (PAS) of AChE have been shown to inhibit the AChE-induced aggregation of the β-amyloid peptide (Alvarez, Opazo, Alarcón, Garrido, & Inestrosa, 1997;Bartolini, Bertucci, Cavrini, & Andrisano, 2003;Inestrosa et al., 1996). It has been shown that AChEIs simultaneously binding to the catalytic active site (CAS) and PAS of AChE (dual-site inhibitors) are responsible for the enhanced binding of gorge-spanning ligands (Gupta & Mohan, 2014). Dual binding site AChEIs have been currently recognized as a new strategy to identify the more efficacious and promising anti-Alzheimer's candidates to positively modify the course of AD (Chierrito et al., 2017;Gupta & Mohan, 2014).
One common approach to design drugs for AD treatment is to modify one known ChEI in order to provide additional pharmacological/biochemical properties other than simple ChE inhibition (Bolognesi et al., 2007). In this sense, tacrine has been widely used to design nontoxic hybrid compounds by combining its potent AChE inhibition with other pharmacological properties (Knapp et al., 1994;Romero, Cacabelos, Oset-Gasque, Samadi, & Marco-Contelles, 2013).
Here we present the synthesis, molecular modeling studies, and biological evaluation of two new tacrine derivatives, a thiosemicarbazone (3) and a semicarbazone (4). The presented results point these classes of compounds as promising leads to the drug design against AD.

Chemicals
Solvents (ethyl alcohol 95% and dioxane) were purchased from VETEC (Brazil); all other reagents were purchased from Merck and Sigma-Aldrich (Brazil). All solvents and chemicals were used without further purification. Reactions were monitored by TLC using DC Alufolien Kieselgel 60 F254 (Merck, Darmstadt, Germany). All NMR measurements were performed at 25ºC on a Varian Premium COMPACT™ 600 MHz (software VNMRJ version 3.2) spectrometer using a 5 mm NMR probe and dimethyl sulfoxide-d6 (DMSOd 6 ) and deuterium oxide (D 2 O). Chemical shifts are given in ppm (δ) with TMS as an internal standard. J values were given in Hertz. Abbreviations for 1 H NMR data quoted are as follows: s (singlet); d (doublet); t (triplet); q (quartet); m (multiplet); bs (broad singlet). IR spectra of the compounds were recorded on a Spectrum 100 spectrometer. All IR and NMR spectra are available in the supplementary information.

Docking studies
The crystallographic structures of human AChE (HuAChE) complexed with donepezil and AChE from Electrophorus electricus (EeAChE) available in the Protein Data Bank (PDB) (http://www.rcsb.org/pdb/home/ home.do) under the codes 4EY7 and 1C2O, respectively, were aligned in the software SPDBViewer (Guex & Peitsch, 1997) and the structure of donepezil from HuAChE was copied to EeAChE in order to provide the model of EeAChE complexed with donepezil to be used in this study. The structures of compounds 3, 4, and tacrine were constructed and had partial charges calculated with Spartan08 ® (Hehre et al., 1999) using the semiempirical molecular orbital method (RM1) (Rocha, Freire, Simas, & Stewart, 2006). Also, the energy minimization was done for all ligands using Merck Molecular Force Field (MMFF) (Halgren, 1996). Docking studies were performed using Molegro Virtual Docker (MVD) software (Thomsen & Christensen, 2006). First, a re-docking study of donepezil over itself inside the model was performed to validate the docking methodology used. After, compounds 3, 4, and tacrine were docked inside EeAChE having donepezil as reference, with different parameters of population size and maximum of interaction. For each ligand, 300 poses were analyzed according to the intermolecular energy values and H-bond interaction residues criteria. The best result obtained for each ligand was chosen to proceed to molecular dynamic (MD) studies.

Molecular dynamic simulations
Before running the MD simulations the ligands had to be parameterized so they could be recognized by the OPLS/AA (Jorgensen, Maxwell, & Tirado-Rives, 1996) force field from the GROMACS program (Abraham et al., 2015). To obtain the parameters and topologies for the referred compounds, the AnteChamber PYthon Parcer InterfacE (ACPYPE) (da Silva & Vranken, 2012) was used. The enzyme/ligand systems were simulated using GROMACS 5.14 package (Abraham et al., 2015) in a cubic box of 1059 nm 3 containing approximately 32,554 TIP4P water molecules. Four steps of minimization (STPR, ST, CG, and LBFG-S) were run, followed by two steps of equilibration (NVT and NPT). After that, in order of allowing the accommodation of the water molecules, 500 ps of MD at 310 K with position restriction (PR) of the enzyme and ligand, were performed, followed by the production step of 20,000 ps of MD simulation at 310 K without any restriction, using 2 fs of integration time and a cut-off of 10 Å for longdistance interactions. Counter ions were added to the systems in order to neutralize the total charge of the enzyme-ligand complexes.
In order to visualize all the steps of the MD simulation, the Visual Molecular Dynamics (VMD) program was used (Humphrey, Dalke, & Schulten, 1996). Plots of random mean square deviation (RMSD), number of H-bonds and mass center distances were done using the Grace program (http://plasma-gate.weizmann.ac.il/Grace/).

Biological evaluation
2.6.1. Cholinesterase inhibitory activity In order to determine the anticholinesterase activity, the modified Ellman's test (Ellman, Courtney, Andres, & Featherstone, 1961) for a 96-wells plate was used. The velocities of substrate hydrolysis by AChE and butyrylcholinesterase (BuChE) as function of sample concentration were evaluated for EeAChE and BuChE from Equine serum (EqBuChE). EeAChE, EqBuChE and Ellman's reagent (DTNB) were prepared in phosphate buffer (100 mM, pH 7.4). Acetylthiocholine iodide (ATCI), and butyrylthiocholine iodide (BTCI) were prepared in distilled water. Stock samples (50 mM) were prepared in dimethyl sulfoxide (DMSO) and appropriately diluted in distilled water to the desired concentrations immediately before use. All solutions were kept on ice during the experiments. All experiments were performed at 37 ± 1°C. All experimental wells received EeAChE (.01 U/mL) or EqBuChE (.05 U/mL), DTNB (.25 mM), and phosphate buffer (control-activity) or sample solutions (.1-100 μM). The mixture was incubated for 10 min. Then, ATCI (.5 mM) or BTCI (1.0 mM) was added to all wells and the plate was read immediately during 5 min in a spectrophotometer (Spectramax 340PC, Molecular Device ® ). The spontaneous hydrolysis of the substrate was evaluated by replacing enzyme for buffer. The solvent (DMSO) was evaluated at the highest concentration (.2%) used in the experiment. All concentrations refer to the final values. The samples were tested in at least five different concentrations. The enzyme activity (absorbance.min −1 ) in a sample solution was determined by comparison with the control (mixture without sample) and expressed as the change in the optic deviation at 412 nm. The values of absorbance.min −1 were calculated by the software Softmax Pro 6.4 ® (Molecular Devices, Sunnyvale, CA). Inhibition values were calculated through non-linear regression with the GraphPadPrism five software (Graph-Pad software ® , San Diego, CA, USA). For each sample, results correspond to average ± standard deviation of two experiments, being each one performed in triplicate. EeAChE, EqBuChE, ATCI, BTCI, and DTNB were purchased from Sigma-Aldrich (Brazil).

Kinetics of enzyme inhibition
The same modified Ellman's test (Ellman et al., 1961) of spectrophotometric analysis was used to determine the type of inhibition. Kinetic parameters were determined using the Lineweaver-Burk double reciprocal method (Lineweaver & Burk, 1934) at increasing concentrations of substrate (.1; .3; .5; 1.0; 1.5; 2.0 mM) below and above K m , keeping a fixed amount of EeAChE in the absence or in the presence of inhibitor. The inhibitors concentrations (3; 10 and 30 μM) were kept close to one which corresponds to the IC 50 and their inhibitory kinetics were evaluated by the Lineweaver and Burk method (Lineweaver & Burk, 1934).

Kinetic study by NMR
All NMR analyses were performed using an NMR method former used in our research group (Soares, Vieira, Delfino, & Figueroa-Villar, 2013) on a Varian Premium COM-PACT TM 600 MHz spectrometer using a 5-mm probe at 25°C. 2 μL of EeAChE 10 μM in phosphate buffer (100% D 2 O, pH 7.4) and in the presence of 1% bovine serum albumin was used. This enzyme solution was mixed with 30 μL of ACh (100 mM in D 2 O), and then it was diluted to 600 μL using phosphate buffer (100 mM, 100% D 2 O, pH 7.4) in the NMR tube (final concentration = 5 mM for ACh and 33 nM for the enzyme). This sample was immediately inserted in the magnet for locking and shimming, allowing for the observation of the first 1 H spectra exactly 5 min after the addition of ACh. The next 1 H spectra were acquired every 5 min with a single scan over 75 min. For testing the new compounds, the same procedure was performed, including the addition of 5 μL of each potential EeAChE inhibitor (final concentration of the inhibitor in the NMR tube = 10 mM) before the addition of ACh. The concentrations of ACh and acetate (Ac) were determined by the integration of the methyl signals (ACh at 2.24 ppm and Ac at 2.16 ppm). All analyses were performed in triplicate.
thiosemicarbazide for obtaining compound 3, and with semicarbazide for getting compound 4 (Scheme 1). The structures of the new synthesized compounds were confirmed by IR and NMR ( 1 H, 13 C, and 2D-NMR). All spectra and the assignments for the molecules and the proposed mechanism for the synthesis can be found in the supplementary information.
The new compounds were synthesized taking into account that the presence of groups like -Br and -NH 2 and also aromatic rings, are important for the interaction with -OH groups and aromatic rings (pi-pi stacking), respectively, in the active site of AChE (Bissantz, Kuhn, & Stahl, 2010;Neto et al., 2017).

Structure optimization and calculation of pharmacokinetic and toxicological properties
The 3D optimized structures of compounds 3 and 4 and their values of energy (au), dipole moment (Debye), molecular weight (amu), area (Å 2 ), polar surface area (PSA) (Å 2 ), volume (Å 3 ), and the electronic pending of carbons and hydrogens, obtained through Spartan08 ® (Hehre et al., 1999) are shown in Figure 1. In the context of the rule of five (Lipinski, 2000(Lipinski, , 2004 PSA values ≤140 and ≤90 Å 2 (Lipinski, 2004;Pajouhesh & Lenz, 2005) mean good oral bio-availability and good penetrability in the blood brain barrier (BBB), respectively. Therefore, the PSA values of 47.72 and 63.27 Å 2 calculated for compounds 3 and 4, respectively, (Figure 1) suggest that these compounds have a good oral bio-availability besides being able to penetrate the BBB and entering the CNS. Table 1 shows the physicochemical properties for compounds 3, 4, and tacrine obtained from OSIRIS Property Explorer software, an online cheminformatics tool (http://www.organic-chemistry.org/prog/peo/). It is known that compounds with cLogP value <5 have a more favorable drug-likeness profile (Pajouhesh & Lenz, 2005;Walters & Murcko, 2002;Zernov, Balakin, Ivaschenko, Savchuk, & Pletnev, 2003) which is observed for compounds 3 and 4. The analysis of donor sites (nOHNH) and hydrogen bonding acceptors (nOH) shows that compounds 3 and 4 (below 10 and 5, respectively) also meet the rule for oral bio-availability and penetration in CNS. Besides, the toxicological profiles show lower toxicity of these compounds if compared to tacrine.

Docking studies
The docking studies were performed on a model of EeAChE complexed with donepezil in order to be as close as possible to the experimental tests. Figure 2 shows the best re-docking result obtained for donepezil Scheme 1. Synthesis of 5,7-dibromo-3,3-dimethyl-3,4-dihydroacridin-1(2H)-one (2), (E)-2-(5,7-dibromo-3,3-dimethyl-3,4-dihydroacridin-1(2H)-ylidene)hydrazinecarbothiomide (3) and (E)-2-(5,7-dibromo-3,3-dimethyl-3,4-dhihydroacridin-1(2H)-ylidene)hydrazinecarboxamide (4).   inside EeAChE. According to literature, the RMSD value of 1.32 Å obtained for this structure (<2.00 Å) is considered enough to validate the docking protocol used (Kontoyianni, McClellan, & Sokol, 2004). The values of energies for enzyme-ligand interaction and H-bonds, as well as the corresponding interacting residues, for the best poses of each ligand, obtained from the docking studies, are shown in Table 2. These results show that compounds 3 and 4 presented better (more negative) values of both intermolecular and H-bond energies, besides interacting with more residues than tacrine. This suggests stabilization in the active site and that both compounds have more affinity to AChE than tacrine. The best poses for compounds 3 and 4 also showed similar interactions with residues of the CAS (Trp86 and Tyr337) and the PAS (Asp74, Tyr124 and Tyr341) of AChE, contributing to the overall stabilization and leading to the lower values of energy compared to tacrine (Johnson & Moore, 2006). This also suggests that these compounds can inhibit AChE through competitive and non-competitive ways.

Molecular dynamics simulations
MD simulations (20 ns) were performed with the best poses of compounds 3, 4, and tacrine inside EeAChE, in order to check the docking results. The results of temporal RMSD, H-bonds and center mass distances, for each simulated complex, are presented in Figures 3-5, respectively.
The temporal RMSD plots (Figure 3) show that all systems were stable after 15 ns of simulations with no observed fluctuations over .250 nm for EeAChE and .125 nm for ligands.
The plots of H-bonds formed during the 20 ns of MD simulations (Figure 4) show that most of the Hbond interactions predicted in the docking studies were also observed during the MD simulations. It is also possible to see that tacrine formed less H-bond interactions during the MD simulations than compounds 3 and 4. This is also in agreement with the docking results.
Plots of variation of the mass center ( Figure 5) show that compounds 3 and 4 approach to more residues of EeAChE than tacrine during the MD simulations. This is an additional corroboration of the more intense interactions of these compounds inside EeAChE. It is also possible to observe in Figure 5 that the distance between tacrine and Trp86 (from the PAS), increases during the MD simulation. However, for compounds 3 and 4 this distance is kept almost constant, with a slight variation at the end of the simulation for compound 3, probably due to some accommodation in consequence of the approaching of residues Asp74, Tyr337 and Tyr341.
The MD studies corroborate docking results for compounds 3 and 4, as they basically presented H-bonds with the same residues in both studies. Besides, both ligands perform H-bonds with the PAS and active site residues, suggesting that they may be gorge-spanning inhibitors of EeAChE. Also, while docking studies pointed hydrophobic interactions with Trp86, MD studies showed that this interaction is kept all over the simulation, added to H-bonds during the time, as shown in the frames of the MD simulations in Figure 6. Therefore, we can tell that Trp86 contributes in both ways to the stabilization of ligands 3 and 4, playing an important role for the binding of these ligands in the PAS of AChE.

Cholinesterase inhibitory activity
The in vitro inhibitory capacities of compounds 3 and 4 over EeAChE and EqBChE, were tested through NMR (Soares et al., 2013) and Ellman's test (Ellman et al.,1961) with tacrine as reference. As the CAS and the PAS of EeAChE and HuAChE are 100% conserved (see Figure S11 in Supplementary material), the results observed for EeAChE can be extrapolated to HuAChE, despite the structural and functional differences between these two enzymes could result in differences in affinity and potency of the inhibitors (Marquis, 1990;Sussman et al., 1991). Compounds 3 and 4 presented low affinity to EqBChE, of around 30% at the concentration of 100 μM. Both compounds inhibited AChE in a concentration-dependent manner (graphics are in the supplementary information) presenting similar inhibitory potency and IC 50 values of 10.30 ± .37 and 8.66 ± .90 μM, respectively. Tacrine is a non-selective inhibitor of both enzymes with major selectivity towards BChE, instead of AChE (Heilbronn, 1961;Pacheco, Palacios-Esquivel, & Moss, 1995). Regarding inhibitory potency, tacrine (IC 50 = 41,41 ± .20 nM) was at least two hundred times more potent than compounds 3 and 4. However one has to remember that our in silico results (Table 1) suggest that, differently from tacrine, compounds   Novel semicarbazones as potencial gorge-spaning ligands of AChE 4107 3 and 4 are non-toxic. On the other hand, the activities of these compounds were similar to rivastigmine in HuAChE, whose IC 50 is 9.12 μM (Jackisch et al., 2009). The low toxicity, their capacity of penetrating the BBB and the selective inhibition of AChE, associated to the known different biological activities of semicarbazones, point to those compounds as potential new gorge-spanning drugs to be further explored in the drug design against AD.

Kinetics of enzyme inhibition
To determine the mechanism of enzyme inhibition, kinetic studies were performed for compounds 3 and 4. The mechanism of inhibition was graphically determined by applying the Lineweaver-Burk plot, using the reciprocal of velocity and substrate concentration (Figure 7(A1) and (A2)). Based on the obtained plot, a mixed type of inhibition was established by both compounds. Using the Lineweaver-Burk secondary plot (Figure 7(B1) and (B2)), Ki values were obtained. Compound 3 presented Ki = 17.25 μM, showing a slightly higher competitive behavior, whereas compound 4 presented Ki = 13.90 μM, showing more uncompetitive behavior. Due to the observed mixed kinetic mechanism, it can be suggested that these compounds can bind to the PAS (Reyes et al., 1997). All these experimental results corroborate the interactions among compounds 3 and 4 and Trp86 observed in the molecular modeling studies. This residue is a well-known and significant residue in the AChE active pocket and the indole ring of Trp86 normally forms a π-π interaction with AChE inhibitors (Shan, Huang, Zhou, Meng, & Li, 2011;Takeuchi & Wagner, 2006).

Kinetic study by NMR
The NMR method (Soares et al., 2013) was used to determine the concentrations of the substrate (ACh) and the product (Ac), which are obtained from direct integration of the corresponding absorption peaks for methyl groups (2.24 ppm for ACh and 2.16 ppm for Ac), as illustrated in Figure 8 for compound 3. The comparison between the integration of the Ac peak in pure AChE (no inhibitor) and in the presence of the inhibitor, allowed the determination of the percentage of inhibition (Table 3). The analyses indicate that compounds 3 and 4 display EeAChE inhibition activity (Table 3), confirming thiosemicarbazones and semicarbazones as potential classes of AChE inhibitors. Compound 3 showed better inhibition of AChE than compound 4, but lower than tacrine, indicating that other similar derivatives may be more effective and that their analogs can be used as prototypes for the preparation of new drugs.

Conclusion
In this work, we presented two new semicarbazones, structurally related to tacrine, able to inhibit AChE. Despite the IC 50 values observed for these compounds were not good enough to point them as new drugs against AD, they presented similar activities to the AD inhibitor rivastigmine and, according to our in silico predictions, showed to be less toxic than tacrine. Also, our molecular modeling results suggest that our compounds are able to bind simultaneously to the CAS and PAS sites of AChE, working as gorge-spanning ligands. Considering our results and that semicarbazones are already recognized by their several biological activities, we believe that this class could be a good option as leads to the design of new drugs against AD, that, after some structural improvements for improving their inhibitory activities, could be able to prevent the AChE-induced aggregation of the β-amyloid peptide, characteristic of the AD evolution, and potentially become more effective drugs against AD.

Supplementary material
The supplementary material for this paper is available online at https://doi.org/10.1080/07391102.2017.1407676. Table 3. Results for the inhibition of EeAChE by NMR (Soares et al., 2013). The graph indicates the percentage of Ac formation in pure AChE sample and in the presence of compounds 3, 4 and tacrine.