Combining Galantamine and Memantine in Multitargeted, New Chemical Entities Potentially Useful in Alzheimer ’ s Disease

: Herein we report on a novel series of multitargeted compounds obtained by linking together galantamine and memantine. The compounds were designed by taking advantage of the crystal structures of acetylcholinesterase (AChE) in complex with galantamine derivatives. Sixteen novel derivatives were synthesized, using spacers of di ﬀ erent lengths and chemical composition. The molecules were then tested as inhibitors of AChE and as binders of the N -methyl- D -aspartate (NMDA) receptor (NMDAR). Some of the new compounds were nanomolar inhibitors of AChE and showed micromolar a ﬃ nities for NMDAR. All compounds were also tested for selectivity toward NMDAR containing the 2B subunit (NR2B). Some of the new derivatives showed a micromolar a ﬃ nity for NR2B. Finally, selected compounds were tested using a cell-based assay to measure their neuroprotective activity. Three of them showed a remarkable neuroprotective pro ﬁ le, inhibiting the NMDA-induced neurotoxicity at subnanomolar concentrations (e.g., 5 , named memagal, IC 50 = 0.28


■ INTRODUCTION
Alzheimer's disease (AD) is a progressive and fatal brain disorder that causes memory loss, steady deterioration of cognition, and dementia. Over 30 million people worldwide are afflicted with the disease. It is estimated that, by 2050, there could be more than 100 million AD patients worldwide. 1 AD dramatically affects the quality of life of the sufferers and their families, and despite massive investments, there are few, if any, effective treatments for AD. The AD pathogenesis involves a complex interplay of genetic and biochemical factors, including an increased production of β-amyloid peptide (amyloid hypothesis) and an increased phosphorylation of the microtubule-associated tau protein (tau hypothesis). 2 AD progression is associated with a significant disruption of several neurotransmitter systems, including the cholinergic, GABAergic, adrenergic, serotoninergic, and glutamatergic systems. 3 Alterations of cholinergic and glutamatergic systems have been extensively investigated, 4 leading to four acetylcholinesterase (AChE) inhibitors, donepezil, galantamine (1), rivastigmine, and tacrine (now discontinued) and one noncompetitive N-methyl-D-aspartate (NMDA) receptor (NMDAR) antagonist, memantine (2, see Figure 1), which have been approved for use in AD patients. The current standard of care for AD is a combination of an AChE inhibitor with 2. 5,6 The rationale for this approach is that the NMDAR antagonist will stop or delay neurodegeneration, while the AChE inhibitor will improve memory and cognition by stimulating the surviving neurons.
The usefulness of NMDAR antagonists in AD patients has been intensively debated over the years. It has long been known that NMDAR activation can be either beneficial, promoting neuronal survival, or detrimental, causing neuronal death. 7 The current consensus is that these opposite effects depend on receptor location: activation of postsynaptic NMDARs is beneficial, whereas activation of extrasynaptic NMDARs promotes neuronal death. This suggests that the latter receptor population should be the target for novel molecules. 7 In this respect, although 2 is far from being optimal, it is the only NMDAR antagonist in clinical use. This is probably because of its preferential action on the extrasynaptic NMDARs. 8 Unlike other AChE inhibitor drugs, 1 has a postulated dual mechanism of action: in addition to inhibiting AChE, 1 has shown a facilitating effect on the nicotinic receptor-mediated transmission via allosteric modulation of the α-subunit of nicotinic receptors. 9 In particular, 1 can enhance synaptic NMDAR activity 10,11 by activating α7 nicotinic receptors located on presynaptic glutamatergic neurons. 12 Working together, 1 and 2 may thus improve neurophysiological responses in AD patients, 13 and their combination might offer a promising therapeutic strategy for AD treatment. 1 can provide the memory-enhancing activity of any AChE inhibitor, but it can also complement the effect of 2. 14 A possible alternative to drug combinations is to create molecules that can hit multiple targets; these are known as multitarget-directed ligands (MTDLs). 15 MTDLs are gaining increasing attention from the drug discovery community. Recent review articles have pointed out that the strategy of targeting two or more proteins with a single compound can provide therapeutic effects superior to those of a selective drug. 15 −18 In addition, MTDLs can offer a number of potential benefits over cocktails or multicomponent drugs. These include the following: (i) reduced uncertainty in clinical development because predicting the pharmacokinetics of a single compound is much easier than with a drug cocktail; (ii) improved efficacy due to the synergistic effect of simultaneously inhibiting multiple targets; (iii) improved safety by decreasing the side effects related to the load of a drug cocktail with a reduced risk of drug−drug interactions; (iv) simplified therapeutic regimen and improved compliance, which is particularly important for elderly AD patients and their caregivers. 19 In the present study, we combined in single new chemical entities the pharmacophoric moieties of 1 and 2 using our dualtarget design strategy 20 schematically reported in Figure 1. The new chimera compounds were initially investigated by docking simulations, taking advantage of the crystallographic structure of AChE in complex with a bivalent derivative of 1. 21 The spacers were designed to allow the new compounds to properly dock into the AChE gorge and to simultaneously contact both sites of the enzyme. In particular, different polymethylene spacers and heteroatoms were chosen (3−8 and 12 −15). Moreover, we sought to evaluate the role of the adamantane amine function through the design of amide ( 16−18) and Nmethyl (9−11) derivatives. In fact, the protonable nitrogen of 2 could establish cation−π interaction with the peripheral anionic site of AChE, thus increasing the affinity for this target. The molecules were then synthesized and tested against AChE and NMDAR. Interestingly, some of the new compounds showed an unexpected ability to interact with the NMDAR containing the 2B subunit of the receptor (hereafter referred to as NR2B), as they were able to displace ifenprodil, a selective NR2B ligand. It has been postulated that NR2B could be the major component of the extrasynaptic NMDAR population. Therefore, the neuroprotective profile of selected molecules was examined using SHSY-5Y cells.

■ RESULTS AND DISCUSSION
In this study, we investigated the feasibility of designing and synthesizing dual-target compounds, which were able to inhibit the AChE enzyme and to bind to the NMDAR. Other studies have previously sought to obtain dual AChE/NMDAR compounds. 22,23 However, for the first time, we combined in single new chemical entities two marketed drugs, 1 and 2 (see Figure 1 for the design strategy). Docking simulations demonstrated that the amine group of 1 was a suitable point to functionalize this fragment, which would afford compounds 3−18. These compounds were then synthesized and biologically characterized.
Computational Design. Docking simulations were carried out to identify the proper length of the methylene linker connecting 1 to 2. These simulations were performed on a model structure of AChE obtained by imposing, on the crystal structure of the human enzyme, the conformation adopted by its Torpedo californica orthologue in complex with a bivalent derivative of 1 (see the Experimental Section for further details). Table 1 reports the results of the docking simulations. Five to six methylenes were the minimal distance that allowed the new hybrid compounds to efficiently contact both sites (the internal and the peripheral anionic sites) of the AChE gorge without excessive strain on the linker. Interestingly, compounds with a linker length of six or seven units were characterized by an optimal balance between binding score and log P. Moreover, compounds with a six-or seven-methylene spacer displayed the most suitable distance between the basic adamantane amine and the indole ring of Trp286 to form a cation−π interaction. It is worth emphasizing that, due to its electronic nature, this specific interaction was not captured by the scoring function. Figure 2 reports the binding mode of 6 at the AChE gorge. The galantamine part of the compound matched almost perfectly the crystallographic conformation, establishing a  Bottegoni et al. 20 In this study, we used the strategy of designing and synthesizing chimera derivatives of (B) galantamine (1) and memantine (2). Suitable spacers of different length and chemical composition were used, affording 16 new chemical entities, 3−18 (see Table 2). hydrophobic interaction with the indole ring of Trp86 and donating a hydrogen bond to the side chain of Glu202. The linker was lodged in a narrow cavity described by Asp74, Tyr124, Phe297, Tyr337, Phe338, and Tyr341. The protonated nitrogen of the adamantane amine formed a cation−π interaction with the indole of Trp286. The adamatane moiety pointed toward the bulk of the solvent and established hydrophobic interaction with Val294, the backbone of Phe295, and the backbone of Gly342.
Chemical Syntheses. 3−18 were obtained following the synthetic approaches depicted in Schemes 1−3. Due to the easier chemical accessibility and availability of 2 with respect to desmethyl galantamine (20), the synthesis of 3−14 and 16−18 began by generating derivatives of 2, 21−23, 39−44, 51−53, and 63−65 as key intermediates, which were then connected to 20 (Schemes 1 and 2). The alkylation of 2 as hydrochloride salt with the appropriate bromo alcohol under basic conditions gave 24−30. Then the amino group was Boc-protected, achieving 31−37, which were treated with tosyl chloride to afford the easily substitutable tosyl-activated alcohols 38−44. 38 was employed for the synthesis of 15 (Scheme 3), whereas 39−44 were subsequently reacted with 20, providing compounds 45− 50. 3−8 were obtained in good yield after carbamate deprotection under mild acidic conditions to avoid degradation of the desired compounds (Scheme 1). 9−11 were synthesized starting from N-methylmemantine (19), which, in turn, was prepared through reductive methylation of the primary amine of 2 with aqueous formaldehyde and zinc dust by means of slight modification of a literature procedure. 24 The synthesis of 9−11 was accomplished by alkylation of 19 to furnish 21−23, and the following activation of the alcohol group was achieved without isolation of the intermediates, which were directly reacted with 20 (Scheme 1). Amides 16−18 were easily prepared by a two-step procedure starting from 2 hydrochloride and the corresponding acylating agents. The resulting intermediates 51−53 were reacted with 20 to furnish 16−18 (Scheme 1). The introduction of heteroatom chains in 12−14 was carried out by selectively activating the appropriate dialcohol 54−56 (Scheme 2). The key intermediates 63−65 were prepared according to the procedure adopted for 38−44 and condensed with 20, affording 66−68. Then HCl-promoted deprotection of the amino group provided final compounds 12−14. Because the above-described strategy failed for 15, a convergent synthesis was performed by condensing 38 with 71, which was derived from nucleophilic substitution between 20 and 69. Finally, HCl-mediated removal of the Boc group furnished the target compound 15 (Scheme 3).
Biological Assays. All the compounds were tested to assess their abilities to inhibit AChE activity and bind to NMDAR. The AChE-inhibiting activities were measured using the classic method of Ellman. 25 Binding to NMDAR and NR2Bcontaining NMDAR was assessed using the assay based on the displacement of [ 3 H]MK-801 and [ 3 H]ifenprodil, respectively. In addition, the inhibition of the NMDA-mediated neurotoxicity (500 μM of NMDA for 6 h) was assessed using the SHSY-5Y neuroblastoma cell line.
Multitargeted Structure−Activity Relationships. All the compounds were initially tested to assess their ability to inhibit the two selected targets, AChE and NMDAR. Table 2 summarizes the biological data generated for compounds 3−18. As reported above, 1 could be suitably modified at the amine function, because substituents at this position could fit into the AChE gorge and contact the peripheral anionic site of the enzyme, identified by the Trp286 residue ( Figure 2). It has been extensively reported in the literature that compounds, which can simultaneously interact with both sites of AChE (internal and peripheral anionic sites), are expected to potently inhibit the enzyme. 26−32 According to these observations, we have discovered quite potent AChE inhibitors. In fact, 5−8, which were able to interact with both sites of AChE (see Figure  2 and Supporting Information), were nanomolar inhibitors of this enzyme. In addition, the introduction of a methyl group on the nitrogen of 2 further improved the AChE-inhibiting potency, with 9 and 11 being the most potent AChE inhibitors of the present series (IC 50 = 1.03 nM and 0.52 nM, respectively). Conversely, replacing methylenes of the spacer with heteroatoms (12−15) had a slight detrimental effect on the AChE-inhibiting potency (docking models of 12−18 in complex with AChE are reported in the Supporting Information). In addition, we also observed that "switching off" the charge on the moiety of 2, by replacing the methylene close to the nitrogen with a carbonyl group, greatly affected the AChE-inhibiting profile. Docking simulations showed that the binding mode of the present series of compounds was quite well conserved (compare Figure 2 and the binding modes reported in Supporting Information), and therefore removal of the charge prevented 16−18 from properly interacting with Trp286 by means of a cation−π interaction, thus decreasing their affinity for the enzyme. Furthermore, reducing the chain length to four methylenes, as in compound 3, prevented it from properly reaching Trp286 and significantly decreased its affinity (IC 50 = 695.9 nM) when compared to 4−8. All these data a Ligand efficiency is expressed as Docking Score/no. of non-hydrogen atoms. b Ligand efficiency lipophilicity. c Distance between the protonated adamantan-1-amine and the center of the indole ring of Trp286. confirmed the binding mode reported in Figure 2 for the present series of chimeras.
As for NMDAR, some of the new compounds were active in the micromolar range (see Table 2), but none of the new derivatives was remarkably potent against this receptor. The  affinity for this receptor appeared to be mainly driven by 2, while the galantamine structure was not detrimental and did not cause any significant drop in affinity. In the MK-801 binding assay, the most potent derivative of the present series was 3, showing a K i value of 2.34 μM. In this case, the chain length did not affect the affinity against NMDAR (compare the potencies of 3−8), while switching off the charge affected the affinities for this receptor (see compounds 16−18), in agreement with previous observations. 33 The introduction of a methyl group on the nitrogen of 2 (9−11) had a detrimental effect on binding to NMDAR. Replacing methylenes of the spacer with heteroatoms (12−15) had a slight detrimental effect on the affinity for NMDAR. The only exception was 15, where a methylene was replaced by an NH group. 15 showed a K i value of 5.41 μM, comparable to that of the most potent derivative of the present series (3, K i = 2.34 μM). As for 15, we can hypothesize that a further protonable nitrogen, present on its spacer, can positively contribute to the recognition between the ligand and the channel that physiologically controls the permeation of bivalent positive cations.
In summary of the profiles against AChE and NMDAR, some of the new derivatives were nanomolar inhibitors of AChE, and micromolar binders of NMDAR. A comment is required here on the affinity-balancing issue. 34 The nanomolar profiles of the new compounds against AChE, along with the micromolar activities against NMDAR, call for a subsequent campaign of optimization to properly equalize the profile of 3−18 against the two targets. However, it should also be noted that, for in vivo efficacy, it is not strictly necessary that the two targets are modulated to the same extent. In the field of multitarget drug discovery, the issue of affinity-balancing is particularly challenging for the following two reasons: (i) increasing the affinity against the first target can decrease affinity against the second, and vice versa; (ii) there is the possibility that the two targets should be engaged to different extents (i.e., diverse target engagement profile) to provide optimal in vivo activity and efficacy. The latter aspect should be carefully considered when selecting two (or more) targets at the very early stages of a multitarget drug discovery program.
There is now compelling evidence that NR2B-containing NMDAR could preferentially contribute to pathologies linked to glutamate overexcitation. 35−37 Therefore, we turned our attention to NR2B, testing all the compounds using the ifenprodil binding assay (see the Experimental Section for details). While most of the compounds were unable to displace ifenprodil (see Table 2), some of the new derivatives (compounds 5−7, 9, 15, and 18) showed K i values in the micromolar range. A comment is required here on the SAR of the present series of compounds against NR2B-containing NMDAR. In fact, we noticed that slight structural modifications  caused a drop in affinity for NR2B-containing NMDAR. To rationalize such a behavior, we carried out docking simulations using the crystal structure of NR2B in complex with ifenprodil (see Experimental Section for details). In the Supporting Information, we report the docking model of NR2B in complex with 5. The compound fitted quite well into the infenprodil binding pocket, establishing relevant hydrophobic and H-bond interactions (for a detailed description of all interactions, see the Supporting Information). The proposed binding mode could explain the good affinity of 5 for NR2B (K i = 4.6 μM), as well as the ability of the compound to displace ifenprodil as experimentally observed. However, the SARs for this receptor subunit remained puzzling and difficult to be properly rationalized on the basis of our docking model. Further studies are required to properly generate meaningful SARs for the present series of derivatives against NR2B-containing NMDAR. These encouraging data prompted us to test some representative compounds using a cell-based assay to assess their neuroprotective profile against NMDA-mediated neurotoxicity. SHSY-5Y cells were treated with 500 μM NMDA for 6 h in the absence or presence of different concentrations of selected compounds, to generate the dose−response curves of inhibition of the NMDA-induced neurotoxicity. Compounds 5, 6, and 18 were remarkably potent in these experimental conditions, as they were able to inhibit NMDA-induced neurotoxicity with subnanomolar potencies. In particular, 6 showed an IC 50 value of 0.21 nM, while 5 (which we named memagal) and 18 showed IC 50 values of 0.28 nM and 0.37 nM, respectively. As negative control, we tested 12 (NR2B K i > 100 μM), which started to inhibit NMDA neurotoxicity at 10 μM.
Other compounds could not be tested up to 10 μM, because they were cytotoxic at this concentration. Under the same conditions, we also tested 2, which affected cell viability at a much higher concentration (718 nM; see Table 2). These cellbased data were extremely promising and not fully explained by the binding potencies (in the micromolar range) of these compounds against NR2B. Interestingly, while a nanomolar antagonist of NMDAR unable to bind to NR2B (i.e., MK-801; see Table 2) inhibited the NMDA-induced toxicity at nanomolar concentration (IC 50 = 7.91 nM), a selective nanomolar ligand of NR2B (ifenprodil) exerted the same effect at a subnanomolar concentration (IC 50 = 0.73 nM). Similarly, we observed that our compounds, which displaced both MK-801 and ifenprodil at micromolar concentrations, were able to inhibit NMDA-induced neurotoxicity in the subnanomolar range. A possible explanation for this behavior could be that nonselective binders, able to displace both MK-801 and ifenprodil, could result in potent neuroprotective agents, due to possible synergistic and/or additive effects.

■ CONCLUSIONS
In AD, there is still a need for drugs that simultaneously stimulate the cholinergic system while suppressing glutamate neurotoxicity, providing molecules endowed with neuroprotective profiles. In light of the recent failures of compounds that block targets belonging to the amyloid and the tau cascades, there is interest in broadening efficacy of anticholinesterase drugs, by contrasting the progressive neurodegeneration responsible for their decline in efficacy. The neuroprotective profile, achieved through modulating NMDARs, can lengthen the time in which anticholinesterase drugs can provide beneficial effects on memory and cognition. With this in mind, we designed and synthesized novel multitargeted compounds obtained by chemically combining 1 and 2, two marketed drugs for the treatment of AD. Our molecules turned out to be remarkably active against AChE with the most potent derivative showing an IC 50 value of 0.52 nM. For NMDAR, we obtained micromolar binders. The most potent derivative (3) showed a K i of 2.32 μM against NMDAR. However, it was slightly more potent against AChE (IC 50 = 695.9 nM). This series of molecules has further highlighted the inherent difficulty of balancing the affinity profiles against diverse targets, when discovering and developing MTDLs. The molecules were then tested against NR2B-containing NMDAR, using the ifenprodil binding assay. A few compounds were active in the micromolar range, with 7 being the most potent binder of NR2B (K i = 2.9 μM). These molecules are the first ever derivatives reported to show a dual AChE/NR2B profile. They could be promising tools for investigating the simultaneous modulation of these two targets in AD. Finally, in light of the NR2B profile, selected derivatives were tested using cellular assays, to investigate the neuroprotective profile after insult by NMDA. While most of the derivatives were inactive or cytotoxic at micromolar concentration, compounds 5, 6, and 18 turned out to be subnanomolar inhibitors of the NMDA-mediated neurotoxicity. The high potencies of these molecules as neuroprotective agents might lie in their ability to nonselectively modulate NMDARs. This intriguing biological profile deserves further investigation as it could open up new avenues for generating leads and drug candidates endowed with optimal anticholinesterase and neuroprotective profiles against AD.

■ EXPERIMENTAL SECTION
Chemical Synthesis. Chemistry. Chemical reagents were purchased from Sigma Aldrich, Fluka (Italy), and TCI-Europe. N-Desmethylgalantamine was purchased from Synfine research. Solvents were RP grade. Dichloromethane was distilled from calcium chloride. Dry dimethylformamide and triethylamine were used as supplied. Nuclear magnetic resonance spectra (NMR) were recorded at 400 MHz on Varian VXR 400 spectrometers and reported in parts per million. All final compounds 3−18 are >95% pure by UPLC-MS analyses. The analyses were performed on a Waters ACQUITY UPLC-MS instrument consisting of an SQD single quadropole mass spectrometer equipped with an electrospray ionization interface and a photodiode array detector. The analyses were performed on an ACQUITY UPLC BEH C18 column (50 × 2.1 mm ID, particle size 1.7 μm) with a VanGuard BEH C18 precolumn (5 × 2.1 mm ID, particle size 1.7 μm). The mobile phases were 10 mM NH 4 OAc at pH 5 adjusted with AcOH (A) and 10 mM NH 4  bromocresol green spray (0.04% in EtOH made blue by NaOH) following heating of the plate. Compounds were named following IUPAC rules as applied by Beilstein-Institute AutoNom (version 2.1), a PC integrated software package for systematic names in organic chemistry. N,3,5-Trimethyladamantan-1-amine (19). Glacial acetic acid (0.382 mL, 6.66 mmol), Zn dust (0.436 g, 6.66 mmol), and 35% aqueous formaldehyde (0.833 mL, 16.65 mmol) were sequentially added to a solution of 2 hydrochloride (0.720 g, 3.33 mmol) in dioxane/H 2 O (1:1). The mixture was stirred at 30°C for 2 days. Then, saturated NH 4 Cl aqueous solution was added (5 mL), and the resulting mixture extracted with CH 2 Cl 2 (3 × 10 mL). The combined organic layers were then washed with brine and dried over Na 2 SO 4 . The crude product was purified by flash chromatography (CH 2 Cl 2 /MeOH/33% aqueous NH 3 9:1:1.8) to achieve 19 as a white solid: 0.480 g (75%). 1  General Procedure for the Synthesis of 54−56. p-TsCl (1.2 equiv) was added dropwise to a mixture of the appropriate diol (1 equiv), Et 3 N (1 equiv), and DMAP (catalytic amount) in dry CH 2 Cl 2 (0.01 M) under N 2 atm, with stirring at 0°C. The stirring was continued at room temperature for 4 h. The reaction was quenched with saturated NH 4 Cl aqueous solution, and the whole was extracted with AcOEt. The extract was washed with brine and dried over Na 2 SO 4 , and, following solvent evaporation, the crude material was purified by flash chromatography (see Supporting Information for details).
General Procedure for the Synthesis of 21−30 and 57− 59. In a pressure tube (Sigma Aldrich, ACE pressure tube) the appropriate bromo alcohol or the activated alcohol 54−56 (1.2 equiv) was added to a vigorously stirred solution of 19 (1 equiv) or 2 hydrochloride (1 equiv) and K 2 CO 3 (2.5 equiv) in DMF (0.2 M). The reaction mixture was stirred at 80°C for 48 h. After evaporation of the solvent, the residue was taken up with water and extracted with CH 2 Cl 2 (3 × 10 mL). The collected organic layers were concentrated, and the crude was purified by flash chromatography (see Supporting Information for details).
General Procedure for the Synthesis of 31−37 and 60− 62. The appropriate secondary amine (1 equiv) was dissolved in THF/H 2 O (2:1) and treated with Na 2 CO 3 (2.5 equiv) and di-tert-butyl dicarbonate (1.5 equiv). The resulting suspension was stirred at room temperature overnight. The solvent was then removed in vacuo, yielding a residue, which was dissolved in water and extracted with AcOEt (3 × 15 mL). The extracts were combined, washed with brine, dried over Na 2 SO 4 , concentrated, and purified by flash chromatography (see Supporting Information for details).
General Procedure for the Synthesis of 38−44 and 63− 65. p-TsCl (1 equiv) was added to a mixture of the appropriate alcohol (1 equiv), Et 3 N (2 equiv), and DMAP (catalytic amount) in dry CH 2 Cl 2 (0.1 M) under N 2 atmosphere, with stirring at 0°C. The stirring was continued at room temperature for 3.5 h. After completion, the reaction was quenched with saturated NH 4 Cl aqueous solution and extracted with AcOEt (3 × 10 mL). Combined extracts were washed with brine, dried over Na 2 SO 4 , and evaporated. The tosylate derivative was then purified by flash chromatography (see Supporting Information for details).
General Procedure for the Synthesis of 45−50 and 66− 68. A mixture of the appropriate activated alcohol (1.2 equiv), 20 (1 equiv) and Et 3 N (2 equiv) in MeCN (0.1 M) was stirred at 80°C for 48 h in a pressure tube. After evaporation of the solvent, the residue was purified by flash chromatography (see Supporting Information for details).
General Procedure for the Synthesis of 3−8 and 12−14. HCl (4 M) in dioxane (1 mL) was carefully added to the appropriate Boc derivative (1 equiv) at 0°C. The resulting solution was stirred at room temperature for 2 h. After removal of the solvent,, the obtained residue was purified by flash chromatography.
Rat Acetylcholinesterase (AChE) Inhibition Assays. The cholinesterase assay method of Ellman 25 was used to determine the in vitro cholinesterase activity. The activity was measured by increase in absorbance at 412 nm due to the yellow color produced from the reaction of acetylthiocholine iodide with the dithiobisnitrobenzoate (DTNB) ion. AChE was prepared from the brain of Wistar rats by homogenizing under a Teflon blender for 10 min in 0.1 M KH 2 PO buffer, pH 8. A stock solution of the enzyme in 0.1 M KH 2 PO buffer (pH 8), containing 0.1% Triton X-100, was kept frozen. Acetylthiocholine iodide and DTNB were prepared daily using 0.1 M KH 2 PO 4 buffer (pH 7). The assay solution consisted of a 0.1 M phosphate buffer, pH 8.0, with the addition of 340 μM DTNB, 0.02 unit/mL of rat AChE, and 550 μM of substrate (acetylthiocholine iodide).
Assay solutions were preincubated for 10 min at 37°C, followed by the addition of substrate. The inhibitory activity of the newly synthesized compounds was assayed by adding different compound concentrations to the reaction mixture described above. All the inhibitors were dissolved in DMSO, whose concentration never exceeded 1% in the final reaction mixture. The rate of the increase in absorbance at 412 nm was followed for 5 min at 37°C with a PerkinElmer spectrophotometer. Assays were carried out with a blank containing all components except AChE to account for the nonenzymatic reaction. The reaction rates were compared, and the percent inhibition due to the presence of tested compounds was calculated. Data are expressed as percentage of inhibition ± SEM of AChE activity, set to 100%. IC 50 values were determined graphically from log concentration−inhibition curves (GraphPad Prism 4.03 software, GraphPad Software Inc.).
Cells were seeded in 96-well microplates (5000 cells/well); on the next day, cells were incubated with 500 μM NMDA, in the absence or presence of different compound concentrations. At least six different concentrations, ranging from 0.01 nM to 100 μM, were tested for standard and newly synthesized compounds. The compounds were added for 5 min prior to addition of NMDA, to determine the inhibition of NMDAmediated cell toxicity.
Within an experiment, each condition was assayed in duplicate or triplicate and each experiment was performed at least three times. The results were calculated by subtracting the mean background from the values obtained from each test condition and were expressed as the percentage of the control (untreated cells). A nonlinear multipurpose curve-fitting program, Prism (GraphPad Software Inc., San Diego, CA), was used for data analysis of cell viability; the IC 50 value represents the compound concentration able to give 50% of NMDA-induced toxicity. Student's t test was used to evaluate whether differences between the experimental groups and the control were statistically significant.
Docking Simulations. Ligand docking simulations were performed on a receptor model generated using ICM3.7 43 based on the crystal structure of the human AChE solved in complex with fasciculin (PDB code: 1B41). 44 Fasciculin was removed from the complex. The protein atoms were assigned the correct atom types according to a modified version of the ECEPP/3 force field. Hydrogen atoms were added. Heavy side chain atoms that were not solved in the crystal structure and assigned occupancy equal to zero (Glu268: cg, cd, oe1, oe2; Gln291: cd, oe1, ne2; Gln369: cd, oe1, ne2; Arg522: cz, nh1, nh2) were added according to the topology of the ICM library of standard residues. None of these side chain atoms turned out to be involved in the definition of the binding pocket (see below), being either far from the binding site or pointing toward the bulk of the solvent.
The side chains of the residues lining the enzyme gorge had to be optimized to allow relaxing of the conformation of Tyr337 and nearby residues in such a way as to bind bulky ligands. For this purpose, the structure of the human enzyme was superimposed and tethered to the crystal structure of TcAChE solved in complex with a dual derivative of 1 (PDB code 1W4L) 21 using the ICM built-in impose_conformation macro. It appeared safe to adopt this procedure because of the high level of conservation between the two orthologues. According to BLAST, TcAChE and hAChE shared 58.9% sequence identity 45 and 75% homology. 46 The superimposition procedure was based on an iterative algorithm 47 that attempts to find the ideal overlapping core between structures. Atom equivalences were established on the fly according to sequence alignment. On account of this approach, a minority of deviating atoms did not affect the overall quality of the superimposition. Then, the tethering procedure pulled hAChE atoms toward the positions of corresponding atoms in TcAChE according to the alignment and by means of harmonic restraints. Finally, tautomeric states of histidines, the positions of asparagine and glutamine side-chain amidic groups, and the side chains in which heavy atoms were added were optimized through a Monte Carlo procedure to improve the hydrogen bonding pattern in the new conformation. Polar hydrogen atoms were also optimized. Ligands were assigned the MMFF force field atom types and charges. 48 The boundaries of the binding pocket were defined by selecting all the residues with at least one side-chain nonhydrogen atom in the range of 5 Å from the crystallographic position of a previously reported dual derivative of 1 (PDB code 1W4L). 21 When considering only the binding site residues, sequence identity between hAChE and TcAChE was 86% and homology 94%. The backbone displacement in the generated hAChE binding pocket conformation measured in terms of RMSD with respect to the original crystal structure was 0.7 Å (1.0 Å including side-chain heavy atoms). Tyr337 was the only amino acid that displayed a significant conformational rearrangement. As expected, the dihedral angles describing the position of the Tyr337 side chain in hAChE (χ1 −133, χ2 −34) ended up closely resembling those of Phe330 in TcAChE (χ1 −137, χ2 −39).
Docking simulations were carried out by means of the Biased Probability Monte Carlo stochastic optimizer as implemented in ICM. The molecular conformation of the system was described by means of internal coordinate variables. The binding site residues were described by precalculated 0.5 Å spacing potential grid maps, representing van der Waals potentials for hydrogens and heavy atoms, electrostatics, hydrophobicity, and hydrogen bonding, respectively. The van der Waals interactions were described by a smoother form of the 6−12 Lennard−Jones potential with the repulsive contribution capped at a cutoff value of 4 kcal/mol. Poses from Monte Carlo sampling were rescored by means of the standard ICM empirical scoring function.
5 was docked into the crystal structure of the heterodimer formed by the amino terminal domains of N1 and N2B subunits of NMDA receptor solved in complex with ifenprodil. 49 The simulation was performed according to the SCARE induced fit docking protocol that has been thoroughly described by Bottegoni and colleagues, 50 and it is only briefly summarized here. SCARE implements the flexibility of the receptor in the simulation generating different local versions of the binding pocket. The protein atoms were assigned the correct atom types according to a modified version of the ECEPP/3 force field. Hydrogen atoms were added. Tautomeric states of histidines, the positions of asparagine and glutamine side-chain amidic groups, and polar hydrogen atoms were optimized. The binding pocket boundaries were defined, selecting the residues with at least one non-hydrogen atom within 5 Å from the crystallographic position of ifenprodil. Then, ifenprodil was removed from the structure. 5 was assigned the MMFF force field atom types and charges. 48 Twelve variants of the receptor binding pocket were generated, systematically omitting pairs of proximal side chains. 5 was docked at each variant according to the already described ICM standard Monte Carlo docking protocol (see above). The best scoring poses were collected and employed to seed independent optimization runs on the receptor, after restoring the original pocket composition and conformation. For each optimized complex, the binding score was calculated. The complex returning the best score was proposed as the putative binding mode.