Design, synthesis and evaluation of novel cinnamic acid derivatives bearing N-benzyl pyridinium moiety as multifunctional cholinesterase inhibitors for Alzheimer’s disease

Abstract A novel family of cinnamic acid derivatives has been developed to be multifunctional cholinesterase inhibitors against AD by fusing N-benzyl pyridinium moiety and different substituted cinnamic acids. In vitro studies showed that most compounds were endowed with a noteworthy ability to inhibit cholinesterase, self-induced Aβ (1–42) aggregation, and to chelate metal ions. Especially, compound 5l showed potent cholinesterase inhibitory activity (IC50, 12.1 nM for eeAChE, 8.6 nM for hAChE, 2.6 μM for eqBuChE and 4.4 μM for hBuChE) and the highest selectivity toward AChE over BuChE. It also showed good inhibition of Aβ (1–42) aggregation (64.7% at 20 μM) and good neuroprotection on PC12 cells against amyloid-induced cell toxicity. Finally, compound 5l could penetrate the BBB, as forecasted by the PAMPA-BBB assay and proved in OF1 mice by ex vivo experiments. Overall, compound 5l seems to be a promising lead compound for the treatment of Alzheimer’s diseases.


Introduction
Alzheimer's disease (AD) is a fatal neurodegenerative disorder that is clinically associated with cognitive impairment, language skill loss and dementia 1 . To date almost 48 million elderly people are affected by AD, and this number is estimated to show unparalleled growth and increase to spread 131.5 million by 2050 2 . Although the etiopathogenesis of AD is unclear, multiple factors such as amyloid-b (Ab) deposits, low levels of acetylcholine (ACh), s-protein aggregation, dyshomeostasis of biometals and oxidative stress, play a vital role in the pathogenesis of AD 3 . The current therapeutic options against AD are composed by one N-methyl-Daspartate (NMDA) receptor antagonist and three acetylcholinesterase (AChE) inhibitors, namely memantine (NMDA), donepezil, rivastigmine and galantamine (AChE) 4 . Nevertheless, these marketed drugs modestly alleviate the symptoms but cannot cure brain damage or stop neuronal degeneration.
AChE and BuChE play a role in cholinergic signaling. According to the cholinergic hypothesis, the decrease in ACh levels results in memory loss and cognitive impairment, and the clinical restoration of cholinergic function is believed to alleviate AD symptoms 5,6 . Furthermore, studies have illustrated that AChE interacts with Ab through the peripheral anionic site (PAS) promoting the formation of steady AChE-Ab complexes, which are more toxic than single Ab peptides 7 . Therefore, dual-site AChE inhibitors may be promising AD drug candidates 8,9 . Among multiple factors, neurotoxic Ab plaques in the brain are a key contributing factor in the pathology of AD. Ab  and Ab  are the key isoforms of Ab peptides. Ab  may aggregate more rapidly and show stronger neuron cytotoxicity than Ab (1-40) does 10,11 . Preventing the formation and accumulation of Ab is a probable therapeutic strategy for AD. The dyshomeostasis of metal ions such as Cu, Fe and Zn is commonly observed in many critical aspects of AD 12 . The Cu 2þ present in the brain at an abnormally high concentration interacts with Ab, leading to the accelerated formation of neurofibrillary tangles (NFTs) and generate reactive oxygen species (ROS), which further induce oxidative impairment in the brain 13 . Moreover, abnormally high levels of redox-active metal ions, such as Fe 2þ and Cu 2þ , in the brain may lead to the formation of ROS 14 .
Notably, compared with the normal levels of metal ions in brain, the subtraction of physiologically essential metals may lead to risk. Consequently, reducing the abnormally high concentration of metals in the brain by chelating the metals is an additional logical approach for AD treatment. Oxidative stress also plays a crucial role in the development of neurodegenerative disorders such as AD. It has been hypothesised that the antioxidant defence system cannot neutralize oxidative species in elderly people 15 . The oxidative stress theory of ageing also suggests that oxidative damage plays an important role in neuronal degeneration 16 . Therefore, drugs that can scavenge oxygen radicals may be used to prevent AD.
Due to the pathological complication of AD, the multi-targetdirected ligand (MTDL) design strategy has been proposed, and a range of compounds have been developed to act on various targets [17][18][19][20][21] . In AD, the effectiveness of multifunctional molecules with two or more distinct pharmacological properties being properly incorporated is higher than that of single-targeted drugs 22 . Therefore, ChEIs with multiple effects, such as Ab disaggregation, neuroprotection and oxidative load reduction may be a significant approach for AD management.
In the search for new cinnamic acid-based derivatives as anti-AD, we have focused on the structure of benzylpyridinium salts ( Figure  1), which may represent a privileged scaffold that can be used to develop new AChE inhibitors [35][36][37] . Docking studies have shown that the N-benzyl pyridinium moiety interacts with the CAS of AChE, and the heterocyclic moiety interacts with the PAS of AChE forming stacking interactions. Proceeding with our researches on natural products with probable use as anti-AD 38-40 , we reasonably combined the cinnamic acid with benzyl-pyridinium to achieve new hybrids that are anticipated to be dual-acting AChE. In this study, all designed compounds were synthesised and evaluated for their biological activities, including their anti-aggregation activity towards Ab, ChE inhibition, antioxidant activity, metal chelating properties, neuroprotective effects against Ab-induced PC12 cell injury and the ability to cross the blood-brain barrier (BBB). Moreover, kinetic and molecular modelling studies were performed to further explore their mechanism of interaction with AChE and Ab. General procedure for the preparation of compounds 3a-b A mixture of compound 1 (10 mmol) and DMAP (10 mmol) in 20 mL anhydrous CH 2 Cl 2 was stirred at 0 C for about 10 min and EDCI (20 mmol) was added to the mixture and stirred at room temperature for 1 h. The compound 2 (10 mmol) was added to the solution, and stirring was continued overnight. The reaction mixture was diluted with H 2 O and extracted with CH 2 Cl 2 . The organic extracts were combined, washed with brine, and dried with anhydrous Na 2 SO 4 , and the solvent was evaporated in vacuo to give the crude product, which was purified by silica gel chromatography with CH 2 Cl 2 :MeOH ¼15:1 as an eluent to afford corresponding target compound as a yellow solid.

Materials
N-(pyridin-4-yl)cinnamamide (3a) Cinnamic acid was reacted with 4-aminopyridine following the general procedure to give the desired product 3a with a yield of 85% General procedure for the preparation of compounds 5a-n Compound 3 (10 mmol), appropriate benzyl chloride (12 mmol) and a catalytic amount of KI in dry acetonitrile (20 mL) was refluxed for 1-2 h. When the reaction was completed as indicated by TLC, the mixture was then concentrated under reduced pressure, and 20 mL of diethyl ether was added. On cooling, the precipitate was filtered and washed with diethyl ether to get the target compounds 5a-n with high yields.

Kinetic analysis of AChE inhibition
To obtain the mechanism of action 5l, reciprocal plots of 1/velocity versus 1/substrate were constructed at different concentrations of the substrate thiocholine iodide (0.05-0.5 mM) by using Ellman's method 35 . Three concentrations of 5l were selected for the studies: 30.0, 15.0 and 7.5 nM for the kinetic analysis of AChE inhibition. The plots were assessed by a weighted least-squares analysis that assumed the variance of velocity (v) to be a constant percentage of v for the entire data set. Slopes of these reciprocal plots were then plotted against the concentration of 5l in a weighted analysis and Ki was determined as the intercept on the negative x-axis. Data analysis was performed with Graph Pad Prism 4.03 software (Graph Pad Software Inc., San Diego, CA).

Molecular modeling studies
Molecular modeling calculations and docking studies were performed using Molecular Operating Environment (MOE) software version 2008.10 (Chemical Computing Group, Montreal, Canada) 43 . The X-ray crystallographic structures of AChE (PDB code 1EVE) and Ab (1-42) (PDB code PDB 1IYT) were obtained from the Protein Data Bank. All water molecules in PDB files were removed and hydrogen atoms were subsequently added to the protein. The compound 5l was built using the builder interface of the MOE program and energy minimized using MMFF94x force field. Then the 5l was docked into the active site of the protein by the "Triangle Matcher" method, which generated poses by aligning the ligand triplet of atoms with the triplet of alpha spheres in cavities of tight atomic packing. The Dock scoring in MOE software was done using ASE scoring function and force field was selected as the refinement method. The best 10 poses of molecules were retained and scored. After docking, the geometry of resulting complex was studied using the MOE's pose viewer utility.
ABTS radical cation scavenging activity assay49 2,2 0 -Azino-bis-2-ethybenz-thiazoline-6-sulfonic acid (ABTS) was dissolved in purified water to a 7 mM concentration. ABTS radical cation (ABTS.þ) was produced by reacting ABTS stock solution with 2.45 mM potassium persulfate (final concentration) and allowing the mixture to stand in the dark at room temperature for at least 18 h before use. The stock solution of ABTS was serially diluted with sodium phosphate buffer (50 mM, pH 7.4) to 100 lM. Trolox and 5a-n at different concentrations (total volume of 50 lL) were added to 150 lL of 100 lM ABTS solution, respectively. After the addition of either trolox or another antioxidant to the ABTS solution, complete mixing of reactants was achieved by bubbling three to four times using plastic pipettes. The optical absorbance of ABTS at 415 nm was measured at 6 min after addition and equilibrated at 30 C. Each individual treatment was repeated for three times and the results of the experiments were compared.
Inhibition of Ab (1-42) self-induced aggregation Inhibition of self-induced Ab (1-42) aggregation was measured using a Thioflavin T (ThT)-binding assay 47 . HFIP pretreated Ab (1-42) samples (Anaspec Inc., Fremont, CA) were resolubilized with a 50 mM phosphate buffer (pH 7.4) to give a 25 lM solution. Each tested compound was firstly prepared in dimethyl sulfoxide (DMSO) at a concentration of 10 mM and 1 lL of each was added to the well of black, opaque Corning 96-well plates such that the final solvent concentration was 10%. The final concentration of each compound was 20 lM and was prepared in independent triplicates. The solvent control was also included. Then, 9 lL of 25 mM Ab (1-42) sample was added to each well and the samples mixed by gentle trapping. Plates were covered to minimize evaporation and incubated in dark at room temperature for 46-48 h with no agitation.
After the incubation period, 200 lL of 5lM ThT in 50 mM glycine-NaOH buffer (pH 8.0) was added to each well. Fluorescence was measured on a SpectraMax M5 (Molecular Devices, Sunnyvale, CA) multi-mode plate reader with excitation and emission wavelengths at 446 nm and 490 nm, respectively. The fluorescence intensities were compared and the percent inhibition due to the presence of the inhibitor was calculated by the following formula: 100 À (IF i /IF o Â100) where IF i and IF o are the fluorescence intensities obtained for Ab  in the presence and in the absence of inhibitor, respectively.

Metal-chelating study
The study of metal chelation was performed in methanol at 298 K using UV-vis spectrophotometer (SHIMADZU UV-2450PC, Kyoto, Japan) with wavelength ranging from 200 to 500 nm. Due to the low solubility in PBS, the compounds were tested in methanol. The absorption spectra of compound 5l (50 lM, final concentration) alone or in the presence of CuSO 4 and FeSO 4 (100 lM, final concentration) for 30 min in methanol were recorded 1 cm-quartz cells. A fixed amount of compound 5l (50 lM) was mixed with growing amounts of Cu 2þ (10-80 lM) and UV spectra were recorded. Variation of absorbance at 363 nm was used to monitor the formation of 5l/Cu 2þ complex 44,45 .

Inhibition of Cu 2þ -induced Ab (1-42) aggregation
For the inhibition of Cu 2þ -induced Ab (1-42) aggregation experiment, the Ab was diluted in 20 lM HEPES (pH 6.6) with 150 lM NaCl. The mixture of the peptide (10 lL, 25 lM, final concentration) with or without copper (10 lL, 25 lM, final concentration) and the test compound (10 lL, 50 lM, final concentration) were incubated at 37 C for 24 h. The 20 lL of the sample was diluted to a final volume of 200 lL with 50 lM glycine-NaOH buffer (pH 8.0) containing ThT (5 lM). The detection method was the same as that of self-induced Ab aggregation experiment 47 .
Cell culture and measurement of cell viability 50 The toxicity effect of the tested compounds on the rat pheochromocytoma (PC12) cells was examined according to previously reported methods. PC 12 cells was obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China) and routinely grown at 37 C in a humidified incubator with 5% CO 2 in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% bovine calf serum, 100 units per mL penicillin, and 100 units per mL of streptomycin. Cells were sub-cultured in 96-well plates at a seeding density of 5000 cells per well and allowed to adhere and grow. When cells reached the required confluence, they were placed into serum-free medium and treated with compound 5l. Twenty-four hours later the survival of cells was determined by MTT assay. Briefly, after incubation with 20 lL of MTT at 37 C for 4 h, living cells containing MTT formazan crystals were solubilized in 200 lL DMSO. The absorbance of each well was measured using a microculture plate reader with a test wavelength of 570 nm and a reference wavelength of 630 nm. Results are expressed as the mean ± SD of three independent experiments. PC 12 cells were grown in RPMI-1640 medium containing 10% (v/v) foetal bovine serum, 100 U penicillin/mL and 100 mg streptomycin/mL under 5% CO 2 at 37 C. The culture media were changed every other day. After pretreatment with different concentrations of test compound (0, 6, 12.5, 25, 50 lM) for 2 h, PC 12 cells were incubated with 25 lM of Ab (1-42) for 24 h. The cell viability was evaluated using MTT assay. Briefly, the cells were treated with 10 lL MTT (5 mg/mL in PBS) for 4 h at 37 C. Then, 200 lL of DMSO was added to dissolve the dark blue formazan crystals formed in intact cells, and the absorbance at 570 nm was detected by a microplate reader. PC12 cells were cultured without test compound or Ab (1-42) as control group and the results were expressed by percentage of control.
In vitro BBB permeation assay Brain penetration of compounds was evaluated using a parallel artificial membrane permeation assay (PAMPA) in a similar manner as described by Di et al. 51 Commercial drugs were purchased from Sigma (St. Louis, MO) and Alfa Aesar (Haverhill, MA). The porcine brain lipid (PBL) was obtained from Avanti Polar Lipids (Alabaster, AL). The donor microplate (PVDF membrane, pore size 0.45 mm) and the acceptor microplate were both from Millipore (Darmstadt, Germany). The 96-well UV plate (COSTAR @ ) was from Corning Incorporated (Harrodsburg, KY). The acceptor 96-well microplate was filled with 300 lL of PBS/EtOH (7:3), and the filter membrane was impregnated with 4 lL of PBL in dodecane (20 mg/mL). Compounds were dissolved in DMSO at 5 mg/mL and diluted 50fold in PBS/EtOH (7:3) to achieve a concentration of 100 mg/mL, 200 lL of which was added to the donor wells. The acceptor filter plate was carefully placed on the donor plate to form a sandwich, which was left undisturbed for 16 h at 25 C. After incubation, the donor plate was carefully removed and the concentration of compound in the acceptor wells was determined using an UV plate reader (Flexstation @ 3). Every sample was analyzed at five wavelengths, in four wells, in at least three independent runs, and the results are given as the mean ± SD. In each experiment, nine quality control standards of known BBB permeability were included to validate the analysis set.
Ex vivo brain penetration 52 Fifteen male OF1 mice, weighing 25 g were used. Animals were housed under controlled light (with a 12-h light/12-h dark cycle, lights on at 7:00 a.m.) at 25 C and proper humidity. Rats were given food and tap water ad libitum.
Donepezil hydrochloride and compound 5l were dissolved 10% DMSO. Animals were divided into three experimental handling groups: mice administered with (i) control (10% DMSO, n ¼ 5); (ii) donepezil (10 lmol/kg, n ¼ 5); (iii) compound 5l (10 lmol/kg, n ¼ 5). Groups of 15 mice were treated i.p. with each compound. The animals were sacrificed 10 min later and brains were quickly removed and frozen on dry ice. Residual AChE activity was determined as previously described by the method of Ellman et al. 41 using brain homogenate preparations as a source of the enzyme: A homogenate of brain samples (10%, w/v) in 0.03 M sodium phosphate buffer (pH 7.4) was prepared. The brain homogenate in volume of 200 lL was mixed with 1% Triton X-100 and centrifuged at 3000 rpm at 4 C for 10 min. Just before analysis of the enzymatic activity, an amount of 100 lL of homogenate was diluted 2.5 times in 0.1 M phosphate-buffered solution (pH 8.0). The reaction took place in a final volume of 300 lL of 0.1 M phosphate-buffered solution (pH 8.0) containing 100 lL of diluted homogenate and 333 lM DTNB solution. To avoid interferences between AChE and BChE activities, 100 lM ISOOMPA (specific BChE inhibitor) was present in the incubation medium. Prior to the addition of the substrate acetylthiocholine, a preincubation period of 5 min was used to eliminate the endogenous ACh present in the homogenates. The reaction was started by the addition of ATCI (450 lM) and the absorbance at 414 nm was evaluated 2 min after the substrate addition. Percent of inhibition was calculated by comparing AChE activity in brain of the drug-treated mice with activity from untreated controls.

Result and discussion
Chemistry As shown in Scheme 1, the target compounds 5a-n were synthesized. First of all, the commercially available cinnamic acid derivatives 1 were activated with 4-dimethylaminopyridine (DMAP) and 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDCI) at 0 C in dichloromethane solutions, and subsequently condensation with compound 2 at room temperature overnight afforded the intermediates 3. Finally, the benzyl pyridinium bromide salts 5 were efficiently obtained by refluxing proper benzyl bromides 4 with the intermediates 3 in dry acetonitrile.

In vitro cholinesterase inhibitory activity and preliminary SAR studies
The activities of compounds 5a-n and the relative compound cinnamic acid against AChE (from electric eel) and BuChE (from equine serum) were examined using the spectrophotometric method of Ellman et al. 41 Donepezil was used as a standard compound for comparison. From Table 1, the novel cinnamic acid derivatives showed high activity towards AChE with IC 50 values in the nanomolar range, and high selectivity for AChE over BuChE, indicating that these derivatives are selective AChE inhibitors. Among the cinnamic acid derivatives, compound 5l (IC 50 ¼12.1 nM) showed the most potent inhibitory activity against AChE, which was 3.3-fold higher than that of donepezil (IC 50 ¼40.2 nM). Moreover, compound 5l exhibited the highest selectivity level towards AChE over BuChE (SI ¼214.9). By contrast, compound 5n exhibited the highest inhibitory activity against BuChE (IC 50 ¼1.9 lM), resulting 2.3-times more potent than that of donepezil (IC 50 ¼4.5 lM). However, the AChE inhibitory activity of cinnamic acid was remarkably low (IC 50 >100 lM), suggesting that the N-benzyl pyridinium moiety is unavoidably required for higher activity. The AChE inhibition by benzyl pyridinium bromide was assayed to be in the micromolar range 40 , which was lower than that of novel cinnamic acid derivatives. This finding suggests that the cinnamic acid skeleton was also essential for AChE activity.
To improve the inhibitory activity of the compounds against ChEs, we introduced substituents with different sizes and electronic properties on the benzene ring of cinnamic acid and on the benzyl group of the N-benzyl pyridinium moiety were varied. The IC 50 value of compounds 5a-n indicated that the presence of methoxy groups at the positions 3 and 4 of the N-benzyl pyridinium moiety increased the ChE activity. For example, the ChE inhibitory activity of compounds 5i-n was higher than that of compounds 5b-g. Moreover, AChE inhibition was also affected by the substituents on the benzyl group. Compared with compound 5a, the introduction of methyl, fluorine and bromine on the meta-, para-position reduced AChE inhibitory activity (compounds 5b-g). For example, compound 5a (IC 50 ¼54.1 nM for AChE) was more potent than compound 5g (IC 50 ¼1450.5 nM for AChE) possessing 4-Br on the benzyl group. By contrast, compared with compound 5h, compounds 5i-n with different substituents on the benzyl group showed an increased AChE inhibitory activity. For example, compound 5l bearing 4-F on the benzyl group exhibited more potent AChE inhibition (11-fold) than did compound 5h. Similar to compound 5a-n for BuChE inhibition, compounds 5f (IC 50 ¼2.5 lM), 5g (IC 50 ¼2.1 lM), 5l (IC 50 ¼2.5 lM) and 5m (IC 50 ¼1.9 lM) that were characterized by a Br substituent showed more potent inhibition. Therefore, we can rationally deduce that the size of a substituent more crucially affects BuChE inhibition than its electronic properties.
Finally compounds 5i-n were selected for evaluation on human AChE. As listed in Table 2, all tested compounds presented IC 50 values in the nanomolar range and were slightly more potent inhibitors for hAChE than for eeAChE. The SARs for hAChE were similar to those drawn for eeAChE inhibition (Table 1). Compound 5l (IC 50 ¼8.6 nM for hAChE) displayed the highest inhibition, which was fourfold higher than that of standard donepezil (IC 50 ¼33.5 nM).

Kinetic study of AChE inhibition
To explore the AChE inhibitory mechanism of action of the cinnamic acid derivatives, the most potent inhibitor, compound 5l, was selected for a kinetic study using Lineweaver-Burk plots 39 . Graphical analysis (Figure 2) revealed both increasing slopes and increasing intercepts with increasing inhibitor concentration. According to this pattern, compound 5l is a mixed-type inhibitor for AChE and might be able to bind to the CAS and the PAS of AChE.

Molecular modeling study of AChE inhibition
To further study the dual-site mode of compound 5l for AChE, a molecular docking study was performed using the software package MOE 2008. 10 42,43 . The X-ray crystal structure of the TcAChE complex with donepezil (hAChE, PDB code 1EVE) was applied to establish the starting model of AChE. As shown in Figure 3, the N-benzyl pyridinium moiety of compound 5l was bound to the CAS of AChE, via aromatic p-p stacking interactions with the phenyl ring from Trp 84 with the ring-to-ring distance of 3.88 Å and the pyridine ring from Phe 330 with the ring-to-ring distance of 349 Å. Moreover, the charged nitrogen of the pyridine ring bound to the CAS was via a cation-p interaction with Trp 84 and Phe 330. The cinnamic acid moiety occupied the PAS formed by Trp 279 and Gln 74. All these results obviously indicated that compound 5l could simultaneously bind to the PAS and CAS of AChE.

Metal-chelating properties of compound 5l
The chelating ability of compound 5l with biometals such as Cu 2þ and Fe 2þ was studied using UV-vis spectroscopy (Figure 4) 44,45 . As shown in the UV-vis spectrum (Figure 4(A)), compound 5l without metal ions showed the absorption maximum at 216 nm and a shoulder at 353 nm. Upon the addition of CuSO 4 , a bathochromic shift in the maximum absorption from 216 nm to 224 nm and in the shoulder from 353 nm to 363 nm occurred, suggesting the formation of the 5l-Cu 2þ complex. When FeSO 4 was added, a red shift in the maximum absorption from 216 nm to 226 nm occurred, suggesting that compound 5l coordinates with Fe 2þ . The complexation ability of compound 5l might be ascribed to the dimethoxy group on the cinnamic acid moiety and to the amide moiety of the compound 46,33 .
To determine the stoichiometry of the 5l-Cu 2þ complex, the molar ratio method was used, for which solutions of compound 5l with accumulative amounts of CuSO 4 were prepared. The UV spectra (Figure 4(B)) showed that the absorbance at 363 nm, related to  the formation of the Cu-5l complex, initially increased at increasing concentrations of CuCl 2 and then plateaued, and the two straight lines intersected at a mole fraction of 1.02. Thus a 1:1 stoichiometry was hypothesised for the 5l-Cu 2þ complex.

Inhibition of self-induced and Cu 2þ -induced Ab (1-42) self-induced aggregation
All compounds tested for ChEs inhibition were also evaluated by a ThT-based fluorometric assay for their ability to inhibit self-induced Ab (1-42) aggregation 47 . Curcumin was used as a reference, because of its known inhibitory activity against Ab (1-42) selfaggregation. The results are gathered in Table 3. Compounds 5a-n exhibited good potencies (46.3-65.6% at 20 lM) compared with Cur (54.6% at 20 lM). Notably, compounds 5k, 5l and 5n (65.6%, 64.7% and 66.3%, respectively, at 20 lM) showed the highest potency. Interestingly, compounds 5h-n with the dimethoxy group on the cinnamic acid moiety exhibited high inhibition against self-induced Ab (1-42) aggregation with inhibition ranging from 53.9 to 66.3% at 20 lM. For example, compound 5l (64.7% at 20 lM) was more potent than that of compound 5e (52.8% at 20 lM). This finding led to the hypothesis that the dimethoxy group might favour Ab aggregation inhibition. However, substituents on the benzyl group (see compounds 5i-n) did not seem to play a role in the inhibition of Ab (1-42) self-aggregation.
As compound 5l showed good inhibitory activity against Ab (1-42) self-aggregation and favourable chelating properties, its ability to inhibit Cu 2þ -induced Ab  aggregation was investigated by a ThT-binding assay 48 . Clioquinol was employed as a reference compound. As shown in Figure 5, the fluorescence of Ab treated with Cu 2þ is 160.3% that of Ab alone, which points out that Cu 2þ hastens Ab aggregation. In comparison, the fluorescence of Ab treated with Cu 2þ and the tested compound decreased dramatically (5l, 68.6% inhibition of Cu 2þ -induced Ab aggregation; CQ, 60.2% inhibition). These results indicated that our compound could inhibit Cu 2þ -induced Ab aggregation by effectively chelating Cu 2þ .

Docking study of compound 5l with Ab (1-42) peptide
To further study the interaction mode with compound 5l for Ab , a molecular docking study was performed using the software package MOE 2008.10 43 . The X-ray crystal structure of the protein Ab structure (PDB 1IYT) from the Protein Data Bank was used. As revealed in Figure 6, the benzene ring of cinnamic acid interacted with the His 6 via a p-p stacking interaction. A hydrogen bond interaction was also observed between the acid amides group of compound 5l and Glu 3. These results indicated that the p-p stacking and the hydrogen bond interactions played crucial roles in the stability of the 5l/Ab (1-42) complex.    self-induced aggregation, the thioflavin-T fluorescence method was used, the mean ± SD of at least three independent experiments and the measurements were carried out in the presence of 20 lM compounds. b Data are expressed as (mmol trolox)/(mmol tested compound). c N means <0.05.

Evaluation of compounds for antioxidant activity
The target compounds were evaluated for their antioxidant efficacy by using ABTS (ferric reducing antioxidant power) assays 49 . Compared with Cur, results showed that tested compounds exhibited no antioxidant ability (Table 3). This finding might be attributed to the absence of the hydroxy group of cinnamic acid moiety.

Cytotoxicity of compound 5l in PC12 cells and neuroprotection against Ab (1-42)-induced toxicity
On the basis of the aforementioned screening results, the potential toxicity effect of compound 5l in PC12 cells was studied 50 . After incubating the cells with compound 5l for 24 h, the cell viability was determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium (MTT) assay. As shown in Figure 7(A), the result revealed that compound 5l at 3-50 l did not significantly affect cell viability, indicating that compound 5l was nontoxic to neuroblastoma PC12 cells.
In the pathology of AD, Ab-induced neuronal cell death is a serious event. To evaluate the neuroprotective effects of compound 5l against Ab-induced neuronal death of PC12 cells, the data were recorded after the cells were exposed to increasing concentrations of compound 5l (6, 12.5, 25 and 50 lM) for 24 h. As can be seen in Figure 7(B), treatment of cells with Ab (1-42) (25 lM) to the growth medium markedly reduced cell viability to 44.2% compared with the untreated cells (control). Compound 5l exhibited neuroprotective effects at concentrations ranging from 6 to 50 lM (6lM: 47.6 ± 4.2%; 12.5 lM: 57.5 ± 1.4%; 25 lM: 63.2 ± 3.5%; 50 lM: 66.6 ± 2.5%). These observations further showed that novel cinnamic acid derivatives bearing the N-benzyl pyridinium moiety can inhibit Ab (1-42) self-aggregation for the treatment of AD.

In vitro BBB permeation assay
For successful central nervous system (CNS) drugs, the first requirement is crossing the BBB to reach brain. The potential ability of these compounds to penetrate into the brain was evaluated using a PAMPA as described by Di et al. 51 Assay validation was completed by comparing the experimental permeabilities of nine commercial drugs with previously reported values (Table 4) 51 . A plot of the experimental data versus bibliographic values gave a    toxicity. Compound 5l was tested for neuroprotective activity against Ab (1-42) toxicity in PC12 cells. Data represent the mean SD of three observations. Ã p < .05 and ÃÃ p < .01 compared to the Ab (1-42)-treated control group.
Compound 5l with good activities against Ab (1-42) aggregation and AChE was selected. The P e value of compound 5l was 2.89 ± 0.37 (CNSþ). It indicated that compound 5l might be able to penetrate the BBB.

Ex vivo brain penetration study
To confirm the brain permeability of compound 5l predicted by the PAMPA, the AChE inhibitory activity was subjected to an ex vivo measurement after intraperitoneal (i.p.) injection of 10 lmol/ kg of compound 5l and donepezil into mice 52 . The mice were sacrificed 10 min after drug administration, and the inhibition (%) of brain AChE activity versus untreated controls was determined. Compared to control, mouse brain AChE activity was found to be inhibited by compound 5l (53.9 ± 2.9%) and donepezil (46.7 ± 3.2%). These results confirmed that compound 5l can cross the BBB in vivo.

Conclusions
In summary, novel cinnamic acid derivatives bearing N-benzyl pyridinium moiety were designed, synthesised and evaluated as multifunctional cholinesterase inhibitors against AD. Among the synthesised compounds, most derivatives displayed potent AChE inhibitory activity and high selectivity for AChE over BuChE. Among them, compound 5l exhibited dual inhibitory potency on AChE and BuChE. The kinetic characterization suggested that compound 5l acted as a mixed-type inhibition, which was consistent with the result of the molecular modelling study. Furthermore, compound 5l showed metal-chelating ability, significant inhibition of Ab aggregation and inhibition of Cu 2þ -induced Ab aggregation, in addition to low neurotoxicity. Compound 5l also showed a neuroprotective effect against Ab (1-42) toxicity in PC12 cells and was proven to penetrate into brain by the PAMPA-BBB assay in vitro and ex vivo experiments. Above all, compound 5l could be deemed as a multifunctional cholinesterase inhibitor and serve as a novel lead compound for treating AD.