Pursuing the Complexity of Alzheimer’s Disease: Discovery of Fluoren-9-Amines as Selective Butyrylcholinesterase Inhibitors and N-Methyl-d-Aspartate Receptor Antagonists

Alzheimer’s disease (AD) is a complex disorder with unknown etiology. Currently, only symptomatic therapy of AD is available, comprising cholinesterase inhibitors and N-methyl-d-aspartate (NMDA) receptor antagonists. Drugs targeting only one pathological condition have generated only limited efficacy. Thus, combining two or more therapeutic interventions into one molecule is believed to provide higher benefit for the treatment of AD. In the presented study, we designed, synthesized, and biologically evaluated 15 novel fluoren-9-amine derivatives. The in silico prediction suggested both the oral availability and permeation through the blood–brain barrier (BBB). An initial assessment of the biological profile included determination of the cholinesterase inhibition and NMDA receptor antagonism at the GluN1/GluN2A and GluN1/GluN2B subunits, along with a low cytotoxicity profile in the CHO-K1 cell line. Interestingly, compounds revealed a selective butyrylcholinesterase (BChE) inhibition pattern with antagonistic activity on the NMDARs. Their interaction with butyrylcholinesterase was elucidated by studying enzyme kinetics for compound 3c in tandem with the in silico docking simulation. The docking study showed the interaction of the tricyclic core of new derivatives with Trp82 within the anionic site of the enzyme in a similar way as the template drug tacrine. From the kinetic analysis, it is apparent that 3c is a competitive inhibitor of BChE.


Introduction
Alzheimer's disease (AD) is a debilitating neurodegenerative disorder that manifests as progressive memory loss leading to dementia [1,2]. AD not only represents a serious health burden, but it also imposes social and economic issues [3][4][5][6]. Worldwide, nearly 50 million people have developed AD, and it is expected that the number will double within the next 20 years [7]. An effective cure able to halt or slow down the disease progression still does not exist, mainly because of our limited knowledge about AD pathophysiology. However, mechanisms resulting in the clinical symptoms of AD are well-established [8,9]. Among them, amyloid-β (Aβ) and tau proteins are considered as major contributors to AD, forming extracellular aggregates [10,11] and intracellular neurofibrillary tangles, respectively [12][13][14]. So far, the therapy is only palliative, either enhancing the cholinergic neurotransmission or modulating the synaptic excitotoxicity via N-methyl-D-aspartate (NMDA) receptors [15][16][17][18]. Oxidative stress [19,20], metal ion imbalance [21], or neuroinflammation [22] are other crucial players in AD pathophysiology.
Currently, AD is treated by acetylcholinesterase (AChE, E.C. 3.1.1.7) inhibitors to restore the physiological levels of acetylcholine (ACh) and the NMDA receptor antagonist, which reduce the excitotoxicity by mitigating the excessive glutamate stimulation of the receptors [23,24]. Donepezil, galantamine, and rivastigmine represent the marketed drugs from the group of AChE inhibitors; memantine acts as a noncompetitive NMDA receptor antagonist, blocking the overstimulation of the respective receptors. Since AChE inhibitors are administered in mild-to-moderate stages of AD, memantine is indicated for severe stages of the disease [25]. It is worth mentioning that the combination of donepezil and memantine into one capsule, known as Namzaric, was approved in 2014. Such a combination offers an improved efficacy compared to single-agent administration with a good pharmacokinetic profile, safety, and tolerability [26,27].
Given the complexity of AD and the high failure rate of single-targeted drug candidates from clinical trials in the last few years, the idea of multitarget directed ligands (MTDLs) has emerged as a new approach to tackle the disease [28][29][30][31]. Accordingly, the design of these molecules is oriented towards the modulation of different pathological pathways simultaneously [29,32]. Recently, we have critically pointed out that not all the combined effects are suitable to be amalgamated into a single molecule to achieve the highest therapeutic effect [29]. Indeed, AD is a long-term condition with progressive brain changes occurring 20 years before the AD outbreak into the symptomatic phase [33]. With respect to the combination of therapeutic targets of interest and to achieve maximum synergy, the ideal drugs should be designed by following the timescale for specific pathological cascades. Within this study, we have focused on the symptomatic stage of AD, in which Aβ plaques and neurofibrillary tangles of paired helical filaments are ubiquitous, and their clearance has no effect on improving cognitive functions. This is well-documented, for instance, by the insufficient efficacy of β-secretase inhibitors in phase III clinical trials [34]. With this in mind, we turned our attention to impaired neurotransmission, which is the most critically affected at the later stages of AD, thus pursuing the two most pronounced systems, namely cholinergic and glutamatergic ones [35,36].
Tacrine (THA), an AChE inhibitor, was the first drug approved for AD treatment in 1993. Hepatotoxicity and gastrointestinal discomfort are the culprits responsible for its withdrawal in 2013 [37]. THA acts not only as AChE inhibitor but, also, via the antagonism of NMDA receptors, which might contribute to its pharmacological effects [38]. Indeed, THA inhibits NMDA receptor responses with high specificity in a concentration-dependent manner with an IC50 = 20 μM at −60 mV and much higher values at positive membrane potentials [39]. Mechanistically, the THA action at NMDA receptors can be classified as reversible, blocking the channel's open state with binding in the proximity of the channel entrance [40,41]. Likewise, other THA derivatives emerged as interesting drugs with a potential applicability for AD treatment due to the pharmacological behavior of both cholinesterases and the NMDA receptor [39,42]. Based on this, our goal was to design a series of novel multipotent agents, which, in one molecule, shows a representative cholinesterase inhibition together with NMDA antagonist activity.
In a recent work, we developed novel molecules building on a fluorene moiety. The compounds are structurally related to other tricyclic congeners like THA or carbazole, known for their cholinesterase inhibition and/or neuroprotective properties (Figure 1) [43][44][45]. We took advantage of the above-mentioned and designed novel molecules substituted at the C9 position by various primary or secondary amines, generating 15 novel compounds (3a-o; Figure 1) as potential cholinesterase inhibitors and NMDA receptor antagonists. Importantly, since several techniques on how to construct MTDLs exist, we applied the so-called merging approach in this study. This approach best corresponds to the drug-likeness of the final molecule compared to the linking or fusing strategies [28,29,32]. Furthermore, the design of the novel family was envisaged in parallel with their physicochemical properties, presuming both oral and central bioavailability. The biological profile includes the assessment of AChE, butyrylcholinesterase (BChE; E.C. 3.1.1.8), and NMDA receptor affinities, with the cytotoxicity and BBB permeation being evaluated as well.

Chemistry
All chemical solvents and reagents were used in the highest available purity without further purification, and they were purchased from Sigma-Aldrich (Prague, Czech Republic) or FluoroChem (Hadfield, UK). The reactions were monitored by thin-layer chromatography (TLC) on silica gel plates (60 F254, Merck, Prague, Czech Republic), and the spots were visualized by ultraviolet light (254 nm). Purification of crude products was carried out using a PuriFlash Gen5 column, 5.250 (Interchim, Montluçon, France) (silica gel 100, 60 Å, 230-400-mesh ASTM, Sigma-Aldrich, Prague, Czech Republic). NMR spectra were recorded in deuterated chloroform (CDCl3) and deuterated dimethyl sulfoxide (DMSO-d6) on a Bruker Avance NEO 500 MHz spectrometer (499.87 MHz for 1H-NMR and 125.71 MHz for 13C-NMR; Vienna, Austria). Chemical shifts are reported in parts per millions (ppm), and spin multiplicities are given as broad singlet (bs), doublet (d), doublet of doublet (dd), triplet (t), doublet of triplet (dt), quartet (q), pentet (p), or multiplet (m). Coupling constants (J) are reported in Hz. Recorded NMR spectra are available in the Supplementary Information. Melting points were measured using an automated melting point recorder M-565 (Büchi, Flawil, Switzerland). The synthesized compounds were analyzed by an LC-MS system consisting of UHLPC Dionex Ultimate 3000 RS coupled with a Q Exactive Plus orbitrap mass spectrometer to obtain high-resolution mass spectra (Thermo Fisher Scientific, Bremen, Germany) (see Supplementary Information). The samples were dissolved in DMSO/methanol 50/50 (v/v). Reverse-phase C18 column Kinetex EVO (Phenomenex, Torrance, CA, USA) was used as a stationary phase, and purified water with 0.1% formic acid (mobile phase A) and LC-MS grade acetonitrile with 0.1% formic acid (mobile phase B) were used as the mobile phases. Gradient elution was used to determine purities and mass spectra. Method started with 5% B for 0.3 min, then the gradient rose to 100% B in the third min and was at 100% B for 0.7 min and then went back to 5% B and was equilibrated for 3.5 min. Total run time of the method was 7.5 min. Column was tempered to 27 °C, the flow of the mobile phase was 0.5 mL/min, and the injection volume was 1 µL. Gradient LC analysis with UV detection (254 nm) confirmed a >97% purity. High-resolution mass spectra were collected from the total ion current in the scan range 105-1000 m/z, with the resolution set to 140,000.

General Procedure for the Preparation of 9-Bromofluorene hydrochlorides (3a-o)
Appropriate primary or secondary amine (Scheme 1) was dissolved in 10 mL of dry MeCN under argon atmosphere and stirred vigorously at 40 °C for 15 min. 9-Bromofluorene (Scheme 1) was dissolved in 5 mL of dry MeCN and added dropwise to the reaction mixture. The reaction was maintained with stirring at 40 °C for 3 h. After cooling, the solvent was evaporated under reduced pressure, and the crude material was purified on a FlashChrom column (eluent dichloromethane/methanol (DCM/MeOH) with 1% of NH3, gradient 95:5 → 90:10) to get the appropriate products as a free base. These were converted into hydrochloride salts by the treatment with 1 mL of concentrated aqueous solution of HCl (35%) in 5 mL of MeOH, starting from 0 °C to room temperature for 1 h. The solvent was removed in vacuo, and the residual water was distilled via azeotropic distillation with absolute EtOH three times. The solid product was washed by ice-cold acetone, resulting in hydrochloride salt as a white solid.    13

In Vitro Anti-ChE Assay
The human AChE/human BChE (hAChE/hBChE) inhibitory activity of the tested derivatives was determined using Ellman's method [46][47][48] and is expressed as IC50, i.e., a concentration that reduces the cholinesterase activity by 50%. The human recombinant BChE and AChE were prepared at the Department of Toxicology and Military Pharmacy . Inhibitor solutions in the concentration range 10 −3 -10 −11 M were prepared, and IC50 values were calculated. Tested compounds were preincubated for 5 min. The reaction was started by the immediate addition of 20 µL of the substrate. The activity was determined by measuring the increase in absorbance at 412 for hAChE/hBChE at 37 °C at 2 min intervals using a multimode microplate reader Synergy 2 (BioTek Instruments, Inc., Winooski, VT, USA). Each concentration was assayed in triplicate. Software GraphPad Prism 5 (San Diego, CA, USA) was used for the statistical data evaluation.

Kinetic Study of hBChE Inhibition
The kinetic study of hBChE was performed by using the above-mentioned modified Ellman's method. The values of Vmax and Km of the Michaelis-Menten kinetics, as well as the value of Ki, were calculated by nonlinear regression from the substrate velocity curves. Linear regression was used for calculation of the Cornish-Bowden plots. All calculations were performed using GraphPad Prism software version 6.07 for Windows (San Diego, CA, USA).

HEK293 Cell Culture and Transfection
Human embryonic kidney 293 (HEK293) cells were cultured in Opti-MEM I containing 5% fetal bovine serum (FBS; both from Thermo Fisher Scientific) [49]. The cells grown in a 24-well plate were put in Opti-MEM I media containing a mixture of 0.9 µL of MATra-A Reagent (IBA) and 900 ng of DNA vectors carrying the human versions of the GluN1-1a (GluN1), GluN2A, or GluN2B subunits and green fluorescent protein (GFP; all vectors were added in equal ratio) [39,50,51]. The cells were placed on a strong magnet plate for 30 min and then trypsinized; resuspended in Opti-MEM I containing 1% FBS, 20-mM MgCl2, and 3-mM kynurenic acid (to inhibit the NMDAR-induced excitotoxicity); and plated on poly-L-lysine-coated glass coverslips. The electrophysiological experiments were executed 24-48 h after transfection [52].

Electrophysiology
Whole-cell patch-clamp recordings were conducted at room temperature on the GFPpositive HEK293 cells using an Axopatch 200B amplifier (Molecular Devices, San Jose, CA, USA); the series resistance compensation was >80% for all cells [39]. The recorded currents were filtered at 2 kHz with an eight-pole low-pass Bessel filter and digitized at 5 kHz with Digidata 1322A and pClamp 10 software (Molecular Devices). The WAS02 application system, which can reach the solution exchange time around the recorded cell of 10-20 ms, was used to apply the extracellular solution (ECS) [53]. The ECS contained (in mM): 160 NaCl, 2.5 KCl, 10 HEPES, 10 glucose, 0.2 EDTA, and 0.7 CaCl2 (pH adjusted to 7.3 with NaOH) and was supplemented with the saturating concentrations of co-agonists glycine (50 µM) and agonist glutamate (1 mM; both from Merck). The stock solutions of 3a-o (10 mM) were prepared freshly before each experiment in dimethylsulfoxide (DMSO; Merck), and the background concentration of DMSO was kept equal in each ECS. The borosilicate glass pipettes with a tip resistance ~4-7 MΩ were made using a P-1000 micropipette puller (Sutter Instrument Co., Novato, CA, USA) and were filled with an intracellular solution (in mM: 125 gluconic acid, 15 CsCl, 5 EGTA, 10 HEPES, 3 MgCl2, 0.5 CaCl2, and 2 ATP-Mg salt (pH adjusted to 7.2 with CsOH). Data were analyzed using SigmaPlot 14.0 (Systat Software, Inc., Chicago, IL, USA), and the dose-response curves were built using Equation (1): where IC50 is the concentration of the compound that produces a 50% inhibition of the glutamate-evoked current, [compound] is the concentration of the studied compound, and h is the apparent Hill coefficient [54].

In Vitro Cell Viability Assessment
Standard MTT assay (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; Sigma Aldrich, Prague, Czech Republic) was used according to the manufacturer's protocols on the CHO-K1 (Chinese hamster ovary, ECACC, Salisbury, UK) and hCMEC/D3 (Sigma Aldrich, St. Louis, MO, USA). The cells were cultured according to recommended conditions and seeded in a density of 8000 cells per well, as described previously [55]. Briefly, the tested compounds were dissolved in dimethyl sulfoxide (DMSO; Sigma Aldrich, St. Louis, MO, USA) and subsequently diluted in the nutrient mixture F-12 ham growth medium (Sigma Aldrich, St. Louis, MO, USA), supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin (both Sigma Aldrich, St. Louis, MO, USA), so that the final concentration of DMSO did not exceed 0.5% (v/v). CHO-K1 cells were exposed to serially diluted tested compounds for 24 h. Then, the medium was replaced by a fresh medium containing 0.5 mg/mL of MTT, and the cells were allowed to produce formazan for another approximately 3 h under surveillance. Thereafter, the medium with MTT was removed, and crystals of formazan were dissolved in DMSO (100 µL/well; PENTA s.r.o., Prague, Czech Republic). Cell viability was assessed spectrophotometrically by the amount of formazan produced. The absorbance was measured at 570 nm with a 650-nm reference wavelength on Synergy HT (BioTek, Winooski, VT, USA). The IC50 value (halfmaximal inhibitory concentration) was calculated from the control-subtracted triplicates using nonlinear regression (four parameters) by GraphPad Prism 7.03 software (GraphPad Software Inc., San Diego, CA, USA) for the CHO-K1 cell line, and for hCMEC/D3 cells, the cell viability was expressed as the % relative to the untreated control. Final IC50 and SEM (standard error of the mean) values were obtained as a mean of three independent measurements.

Determination of In Vitro BBB Permeation
The D3 assay evaluates the ability of compounds to diffuse from the donor compartment through the D3/hCMEC cell membrane into the acceptor compartment. The D3 cells were seeded on a PET membrane (area 1.12 cm 2 ) with 3-µm pores of the 12-well plates with 12-mm inserts. The tested compounds were dissolved in DMSO and then diluted with Opti-MEM to reach final a concentration of 30 µM for 3m and 50 µM for others; the concentration of DMSO did not exceed 0.5% (v/v). The donor solution (750 µL) was added to the donor compartment (insert), and the same volume of Opti-MEM was added into the acceptor. The concentration of the drug in both compartments was measured by UV-VIS spectrophotometry in 5, 15, 30, and 60 min of incubation in triplicates. The apparent permeability coefficient (Papp) was calculated from the concentration ratios. Tightness of the D3 monolayer was conducted by the fluorescein isothiocyanate (FITC) test after each experiment. The accepted values of FITC (fluorescein isothiocyanate) in 0.4 mg/mL in the acceptor compartment after 30 min of incubation must not exceed 1.5% of the initial donor concentration.
A: area of the well/cell monolayer, dC/dt: amount in the receiver compartment in given time, Vr: volume of the receiver compartment, and C0: the initial concentration of tested compounds.

Molecular Modeling Studies
The model of hBChE (Protein Data Bank (PDB) ID: 4BDS, resolution: 2.10 Å) was downloaded from the RCSB Protein Data Bank (https://rcsb.org) and prepared for flexible molecular docking by MGL Tools [56,57]. The preparation of these receptors involved removal of the inhibitor, water molecules and nonbonded co-crystalized compounds, and the addition of polar hydrogens. Default Gasteiger charges were assigned to all atoms. Flexible parts of the enzymes were set based on previous experiences [58,59]. The rotatable bonds in the flexible residues were detected automatically by AutoDock Tools 1.5.

In Silico Pharmacokinetics and Drug-Likeness Prediction
SwissADME, web tool was used to predict gastrointestinal absorption, BBB permeation and bioavailability [62]. Physicochemical properties (pKa, ClogP, HBA, HBD, and TPSA) were predicted by MarvinSketch 20.4.0, ChemAxon Ltd. (Budapest, Hungary; see Table S1, Supplementary Information). BBB scores were calculated by the BBB calculator available at the ACS website [63].

In Silico Prediction of the CNS and Oral Availability
All the newly designed compounds were initially screened in silico to predict their peroral and CNS availability. Indeed, both of these features are essential in the drug discovery process of small molecules as potential anti-AD therapeutics and should be estimated before synthesis. To this end, the in silico pharmacokinetics; drug-likeness; and ADME (absorption, distribution, metabolism, and excretion) prediction was applied to compounds 3a-o using a web-based tool SwissADME [62,64,65]. THA and memantine were used as references with known CNS statuses and pharmacokinetic profiles [42,66,67]. Data are presented in Figures 2 and 3 and Tables 1 and S1.    [68], and e Bio. Score = bioavailability score [69].
The pink middle area (Figures 2 and 3) represents the optimum range for bioavailability. It is a sum of the lipophilicity (LIPO), size, polarity (POLAR), solubility (INSOLU), saturation (INSATU), and flexibility (FLEX). The red lines show the predicted state for the designed compounds [62,64,65]. The calculated parameters for all the compounds, except for 3b and 3h, are situated inside the pink area, suggesting their high oral bioavailability [69]. A positive correlation for the bioavailability was also supported by the BOILED-Egg method [70], proposing high gastrointestinal adsorption and penetration through the BBB [70]. The BBB score, a new predictive model for CNS availability, was also calculated for all the derivatives fitting the range between 5 and 6, indicating a high probability to permeate through the BBB (estimated physicochemical parameters are shown in Table S1, Supplementary Information) [63]. Besides, all compounds also fulfill the drug-likeness rules of pharmacological models defined by pharmaceutical companies like Veber's (GSK) [71], Lipinski's (Pfizer) [72], Egan's (Pharmacopeia) [73], Ghose's (Amgen) [74], and Muegge's (Bayer) [75] (Table S2, Supplementary Information). In summary, derivatives 3a-o unambiguously show high predictive peroral bioavailability, BBB permeation, and acceptable pharmacokinetic profiles, making the compounds prospective candidates for the treatment of CNS disorders, including AD.

Synthesis
Fluoren-9-amines were prepared in a one-step procedure from commercially available 9-bromofluorene (1) with an excess of various primary or secondary amines (2a-o) in dry acetonitrile (Scheme 1). This reaction afforded the formation of fluoren-9-amines (3ao) in high yields (80-95%). The free bases were then converted to the respective hydrochloride salts. The final products were characterized by 1 H, 13 C-NMR spectra, melting points, and HRMS analysis. LC analysis confirmed the purity for all the derivatives > 97% (Supplementary Information).

Evaluation of Cholinesterase Inhibitory Activity
Fluoren-9-amines were tested for their inhibitory potential against human AChE (hAChE) and human BChE (hBChE) enzymes using the modified spectrophotometric method of Ellman et al. [46,48,76]. THA was used as a reference compound. The IC50 values of all tested compounds and their selectivity indexes (SI) for hBChE are summarized in Table 2. All tested derivatives showed a preferential inhibition of hBChE (usually in the single-digit micromolar range), whereas only compounds 3c and 3m showed weak inhibitory activity for hAChE at 13.98 and 49.91 µM, respectively. The hBChE inhibition ranged between 0.47-54.65 µM. According to the hBChE inhibition potency, the top-ranked derivative from all the derivatives under the study was 3c (IC50 = 0.47 µM), being 20 times less active compared to THA. However, the relatively unique selectivity to BChE is of immense interest. Indeed, the selectivity profile for the ChE drugs potentially useful in AD treatment is extensively discussed [77]. The currently used ChE inhibitors are AChEselective (donepezil) or possess more or less nonselective patterns of inhibition (rivastigmine and galantamine) [25]. It was proven that the levels of BChE in the brain of AD patients are elevated with age, while the levels of AChE are suppressed. This finding underlines the importance of selective-BChE inhibitors in the treatment of the later stages of AD [77,78].

Kinetic Study of hBChE Inhibition
To determine the inhibition pattern of 3c, a kinetic study was performed to elucidate the interaction mechanism of the compound with hBChE. The kinetics of the inhibition were revealed from velocity curves measured at several concentrations of 3c and the hBChE substrate butyrylthiocholine. The type of enzyme inhibition and appropriate affinity parameter (Ki) were determined using nonlinear regression analysis. The results for each type of model of inhibition (competitive, noncompetitive, uncompetitive, and mixed) were compared by the sum-of-squares F-test. A statistical analysis showed a competitive type of inhibition for hBChE (p < 0.05) that was in line with the Cornish-Bowden plot used for visualization of the obtained data (Figure 4).
.  The parallel lines indicate a competitive inhibition of hBChE by 3c, which means a reversible binding mode to the active site of the enzyme. With the increasing concentration of the inhibitor, the apparent Vmax remained unchanged and Km increased. A Ki value of 173 ± 25 nM was determined for 3c on hBChE.

Evaluation of Antagonist Activity towards the NMDA Receptor
Further, we aimed to examine the effects of all synthesized compounds (3a-o) at both the GluN1/GluN2A and GluN1/GluN2B receptors. First, we held the transfected cells at a membrane potential of −60 mV, and we assessed the relative inhibitory effect of each compound at a 10-µM concentration upon its application with the (co-)agonists (Figure 4). These experiments showed that all studied compounds exhibited an inhibitory effect ranging from ~7% to ~52% at both studied NMDAR combinations (Table 3). Interestingly, we revealed that 3e was the most potent compound at the GluN1/GluN2A receptors, while 3m was at the GluN1/GluN2B receptors. Subsequently, we generated concentrationresponse curves for 3e and 3m (1-300 µM) for both the GluN1/GluN2A and GluN1/GluN2B receptors at the negative (−60 mV) and positive (40 mV) membrane potentials ( Figure 5). Our analysis showed that both compounds exhibited a more profound inhibitory effect at the negative membrane potential (IC50 values ranged from ~9 µM to ~15 µM), while they were much less potent at the positive membrane potential (IC50 values ranged from ~83 µM to ~221 µM); the results are summarized in Table 4. Together, our data show that both 3e and 3m act as potent voltage-dependent inhibitors of the NMDARs, suggesting that they act as open-channel blockers of NMDARs. Interestingly, the IC50 values for both 3e and 3m at the GluN1/GluN2A and GluN1/GluN2B receptors were in a similar range to those previously obtained with THA (Table 4) [39]. However, both 3e and 3m were less potent than memantine under the studied conditions (Table 4). Table 3. The inhibitory effect of the derivatives 3a-o at the GluN1/GluN2A and GluN1/GluN2B receptors.

Compound
GluN1/GluN2A GluN1/GluN2B RI a (%) ± SEM n RI a (%) ± SEM n 3a 33 The relative inhibition (RI) was calculated as the ratio of the steady-state currents in the presence and absence of 10 µM of the compound (multiplied by 100) at the membrane potential of −60 mV (see text for more details). Data are shown as mean ± SEM; n = number of recorded cells (see Figure 5). ; the values of the IC50 (in µM), Hill coefficient (h), and numbers of analyzed cells (n) obtained from the membrane potentials of −60 mV or 40 mV are shown as mean ± SEM. Dose-response relationships of the inhibitory effects of 3e or 3m at the GluN1/GluN2A and GluN1/GluN2B receptors (see Figure 6). b THA values were published by our group recently [39]. Mem corresponds to memantine.  Table 3).  Table 4.

In Vitro Cell Viability Assessment
The in vitro cytotoxicity evaluation on mammalian cells served as a preliminary toxicity indicator for the newly developed compounds (Table 5). In our previous work, we observed a correlation between the calculated logP (ClogP) values (MarvinSketch software, v. 18.24.0; Table 5) and the toxicity; however, our results within the presented study showed no clear relationship [80]. For instance, one of the most lipophilic compounds (3f) yielded as the least toxic, whereas other more hydrophilic fluoren-9-amines were regarded as low-toxic (e.g., 3g as the most hydrophilic compound) agents as well. The least cytotoxic compound can be classified as 3f with an attached piperidine moiety. At the pole separation, the most cytotoxic agents were 3c, 3m, and 3e. Importantly, some of the new derivatives were less cytotoxic than THA.

In Vitro BBB Permeation
The top-ranked BChE inhibitors and NMDA receptor antagonists were selected (i.e., 3e, 3n, 3c, and 3m) to evaluate their ability to cross the BBB. For this reason, we used the human brain microvascular endothelial cell line hCMEC/D3 as a suitable model for the normal human BBB [82]. Before that, we investigated whether the applied concentration (50 µM) of the tested compound negatively influenced the cell viability in the monolayer. Despite the relatively low toxicity of the CHO-K1 model, we investigated the effects on the cell viability of the hCMEC/D3 line using MTT at 100 and 50 µM (Table S3). We found that all tested compounds at 50 µM did not substantially reduce the cell viability (>90% of viable cells), except 3m, which reduced the cell viability to approx. 82% at 50 µM. Due to this fact, a concentration of 30 µM was used for the BBB permeation test for 3m.
To investigate the potential of the BBB permeation, we used a panel of reference drugs for which the BBB permeation in vivo is known and correlated the Papp values to them. Thus, our measurements predict that the compounds 3c, 3e, and 3n have a low potential to passively pass the BBB. On the other hand, the Papp value for 3m corresponds to those of standard drugs with high CNS permeability (Table 6).

In Silico Studies
To better understand the inhibition mechanism between hBChE with 3c ( Figure 7A,B) and 3n ( Figure 7C,D), respectively, as the top-ranked inhibitors of the respective enzyme, we performed a molecular modeling simulation. The crystal structure of the THA-hBChE complex ( Figure 7E,F; PDB ID: 4BDS) was used for comparative purposes [56]. In line with the compounds' designs, both of the inspected fluoren-9-amines 3c and 3n adopted a similar arrangement to that of THA within the hBChE active site. The crucial observation is that the tricyclic cores of both new molecules interact with Trp82 via parallel π-π stacking in the anionic site of the enzyme. The protonated amino group enabled hydrogen bond formation to carboxylic oxygen from Asp70 (3c and 3n) and the hydroxyl group from Tyr332 (3n). In the case of THA, this type of interaction is replaced by two water bridges with the primary amino group. From the catalytic triad residues (His438, Glu197, and Ser198), only His438 is involved in hydrophobic contact with 3c and 3n, whereas THA is involved in direct hydrogen contact with this residue. The docking results also highlighted the importance of the attached appendages. Both aliphatic parts, either the cyclohexyl (3c) or iso-butyl (3n) moiety, are engaged in hydrophobic contacts with the Tyr332, Phe329, and Ala328 residues. The cyclohexyl moiety of 3c seems to deliver another hydrophobic contact to Pro285 and Gly78 that presumably mediates a slightly higher inhibition potency. Contrary to THA anchoring, missing cation-π contacts in the 3c-and 3n-hBChE complexes can be denoted as the culprit responsible for the slightly decreased affinity of these two agents.  [56]. Ligands are displayed in green (A), salmon (C), and brown (E) for 3c, 3n, and THA, respectively. Important amino acid residues involved in the enzyme-ligand interaction are shown in dark-blue, and the catalytic triad is rendered in yellow. Distances are colored as dashed lines, and the distance is measured in Å (A,C,E). 3D figures were created by using PyMOL 2.0.6 viewer. 2D diagrams were generated with Maestro 12.3 (Schrödinger Release, Schrödinger, LLC., New York, NY, USA).

Conclusions
Since the cholinergic hypothesis was first postulated [15], five decades of intensive research have generated a few marketed drugs for symptomatic relief only. The proper and in-depth elucidation of the disease biology imposes challenging tasks that remain to be resolved by delivering the drugs in the right stage of the disorder [29]. So far, the attempts to approve any drugs have continuously failed, with memantine receiving the final approval in 2003. A glimmer of hope emerged in 2019 when sodium oligomannate (GV-971) was approved in China, acting in a completely different way to other marketed drugs. Indeed, GV-971 suppresses gut dysbiosis, mitigates neuroinflammation, and reverses cognition impairment [83]. Designing MTDLs is an interesting approach that has generated several preclinical candidates for the therapy of AD. However, most of these newly developed agents suffer from limitations, as disclosed recently [29]. Our study was inspired by the appealing combination of donepezil and memantine into one capsule that resulted in a higher therapeutic benefit than the single-agent administrations [27]. Herewith, pursuing the symptomatic stage of AD by modulating the cholinergic and glutamatergic systems, we developed a series of 15 molecules with fluorene tricyclic scaffolds substituted with secondary or tertiary amines at C9. Prior to the synthesis, we took the compound's predicted physicochemical parameters and calculated their oral and CNS availability. Indeed, all the compounds adhered to these criteria. Under in vitro conditions, we demonstrated that the compounds can interact with both NMDAR and BChE while having only a marginal effect on AChE. We also showed that they have a relatively low cytotoxic profile, as determined using CHO-K1 cell lines, and for one compound (3m) out of the four selected candidates, a BBB permeation ability was confirmed. The obtained data herein warrant future research with this class of compounds; however, we may state that we found low-toxic BChE-selective compounds with antagonistic activity on the NMDARs that potentially permeate the BBB.