Combining Chalcones with Donepezil to Inhibit Both Cholinesterases and Aβ Fibril Assembly

The fact that the number of people with Alzheimer’s disease is increasing, combined with the limited availability of drugs for its treatment, emphasize the need for the development of novel effective therapeutics for treating this brain disorder. Herein, we focus on generating 12 chalcone-donepezil hybrids, with the goal of simultaneously targeting amyloid-β (Aβ) peptides as well as cholinesterases (i.e., acetylcholinesterase (AChE) and butyrylcholinesterase (BChE)). We present the design, synthesis, and biochemical evaluation of these two series of novel 1,3-chalcone-donepezil (15a–15f) or 1,4-chalcone-donepezil (16a–16f) hybrids. We evaluate the relationship between their structures and their ability to inhibit AChE/BChE activity as well as their ability to bind Aβ peptides. We show that several of these novel chalcone-donepezil hybrids can successfully inhibit AChE/BChE as well as the assembly of N-biotinylated Aβ(1–42) oligomers. We also demonstrate that the Aβ binding site of these hybrids differs from that of Pittsburgh Compound B (PIB).


Introduction
Alzheimer's disease (AD) is a progressive neurodegenerative disorder that causes brain cells to waste away, resulting in deterioration of memory, cognitive, and executive functions [1][2][3][4][5][6]. Even though the estimates vary, there are as many as 5.5 million Americans age 60 and older with AD [7]. There are no medications available that cure AD or stop the disease progression in the brain. In the advanced stages, complications from the disease can cause dehydration, malnutrition, or infections that ultimately may lead to death [8][9][10][11][12][13]. Since increasing age is the most common risk factor for the development of AD, barring the implementation of an effective treatment or prevention of AD, the number of people with AD will increase significantly if the current population lifespan trends continue. The few medications that are approved for AD (donepezil, tacrine, rivastigmine, galantamine (acetylcholinesterase inhibitors (AChEIs)) and memantine (N-methyl-d-aspartate (NMDA) receptor antagonist) provide only limited symptomatic relief rather than affecting disease progression or providing a cure [14][15][16][17]. Since the number of AD-related drugs used in AD patients is limited, there is an urgent need for novel therapeutic candidates to treat this brain disorder. Scheme 1. Synthetic schemes for the preparation of (A) 1,3-chalcones 4a-4f and (B) 1,4-chalcones 7a-7f.
We next synthesized the donepezil nucleophile 14 in five steps (Scheme 2) consisting of the hydrogenation of ferulic acid 8 in the presence of Pd/C, followed by cyclization in the presence of methanesulfonic acid to result in formation of compound 10 in a 67% yield. Since the condensation of compound 10 with aldehyde 12 was found to be problematic, we had to protect the free hydroxyl group of compound 10 with a tert-butyldimethylsilyl (TBDMS) group to afford the protected compound 11 in 90% yield. Compound 11 was then successfully condensed with aldehyde 12 in the presence of KOH to give compound 13 in 65% yield. The controlled deactivation of the palladium catalyst with thioanisole was essential for selective reduction of the double bond in the presence of ketone and benzyl groups in compound 13 to provide the 6-O-desmethyl donepezil adduct 14 in 94% yield. With the donepezil nucleophile 14 in hand, we finally reacted its free hydroxyl group with two sets of electrophilic alkylated 1,3-and 1,4-chalcones, 4a-4f and 7a-7f, to afford the desired 1,3-and 1,4-chalcone-donepezil hybrids 15a-15f and 16a-16f in 48% to 100% and 37% to 100% yields, respectively. Scheme 1. Synthetic schemes for the preparation of (A) 1,3-chalcones 4a-4f and (B) 1,4-chalcones 7a-7f.
We next synthesized the donepezil nucleophile 14 in five steps (Scheme 2) consisting of the hydrogenation of ferulic acid 8 in the presence of Pd/C, followed by cyclization in the presence of methanesulfonic acid to result in formation of compound 10 in a 67% yield. Since the condensation of compound 10 with aldehyde 12 was found to be problematic, we had to protect the free hydroxyl group of compound 10 with a tert-butyldimethylsilyl (TBDMS) group to afford the protected compound 11 in 90% yield. Compound 11 was then successfully condensed with aldehyde 12 in the presence of KOH to give compound 13 in 65% yield. The controlled deactivation of the palladium catalyst with thioanisole was essential for selective reduction of the double bond in the presence of ketone and benzyl groups in compound 13 to provide the 6-O-desmethyl donepezil adduct 14 in 94% yield. With the donepezil nucleophile 14 in hand, we finally reacted its free hydroxyl group with two sets of electrophilic alkylated 1,3-and 1,4-chalcones, 4a-4f and 7a-7f, to afford the desired 1,3-and 1,4-chalcone-donepezil hybrids 15a-15f and 16a-16f in 48% to 100% and 37% to 100% yields, respectively.
With the goal of investigating the importance of the covalent linkage between the chalcones and donepezil, we synthesized chalcone and donepezil fragments with a C 2 linker (Scheme 3). The parent 1,3-chalcone 3 was reacted with bromoethanol in the presence of K 2 CO 3 to give compound 17 in 24% yield. In the case of the donepezil fragment, the hydroxyl group of bromoethanol was protected with dihydropyran to give compound 18 in 62% yield. This molecule was subjected to a nucleophilic substitution reaction with the donepezil nucleophile 14 to yield 19, which was in turn subjected to the removal of the tetrahydropyran group to give compound 20 in 44% yield.
In order to understand the utility of the hybrids, based on their IC 50 values against EeAChE and Ef BChE, we determined their selectivity index (SI) ( Table 1). For all but two hybrids, 15a and 15b, EeAChE was inhibited equally to >60-fold better than Ef BChE. In the case of 15a and 15b, the inhibition of Ef BChE was 51-fold and 22-fold better than that of EeAChE, respectively.

3 H-PIB Binding Studies
Previously, we demonstrated that various chalcones can displace 3 H-PIB from the Alzheimer's disease PIB binding complex (ADPBC) purified from AD brains, as well as from more widely available model fibrils assembled from synthetic Aβ  and Aβ (1-42) peptides (indicated herein as F 40 and F 42 , respectively) [59]. When studied for their ability to displace 3 H-PIB from F 40 (Figure 1 turquoise bars), F 42 (brown bars), and ADPBC (orange bars), we found that 1,3-and 1,4-chalcone-donepezil hybrids did not displace 3 H-PIB as efficiently as chalcones 3 and 6. Even though there seemed to be no direct correlation between the linker length of the hybrids and their activity, hybrids with shorter linkers appeared to better displace 3 H-PIB from F 40 and F 42 (i.e., compounds 15a and 15b displayed better binding to F 40 , F 42 , and ADPBC than 15c, 15d, 15e, and 15f; compound 16a displayed better affinity for F 40 and F 42 than 16b, 16c, 16d, 16e, and 16f). Interestingly, hybrids 16e and 16f displayed ADPBC affinity comparable to that of 16a. These results indicated that the chalcone-donepezil hybrids probably bind to the fibrils at a different location than 3 H-PIB does in the AD brain. Interestingly, during the course of these experiments, we observed that hybrids 15a-15f and 16a-16f enhanced binding to F 40 . We speculate that when the hybrids bind to F 40 , which is known to be more malleable than F 42 , the conformation of the F 40 peptide/fibril changes in a way that allow for additional PIB binding. Investigation of this phenomenon is beyond the scope of the current study.
Molecules 2019, 24, x 7 of 31 enhanced binding to F40. We speculate that when the hybrids bind to F40, which is known to be more malleable than F42, the conformation of the F40 peptide/fibril changes in a way that allow for additional PIB binding. Investigation of this phenomenon is beyond the scope of the current study. To confirm the results obtained from the 3 H-PIB binding competition studies, we titrated the 1,3and 1,4-chalcone-donepezil hybrids against 3 H-PIB bound to F40, F42, and ADPBC fibrils. The experiments were performed in duplicate for F42 and ADPBC and as a single experiment for F40. Even though the single independent experiment may not be sufficient to determine the exact value of inhibition, we were able to observe a clear trend, which is sufficient at this stage of investigation. The data obtained by titration assays (Figure 2) were completely in accord with those displayed in Figure  1 for both the 1,3-and 1,4-chalcone-donepezil hybrids. was used as a positive control. F 40 = Aβ (1-40) fibrils; F 42 = Aβ (1-42) fibrils; ADPBC = PIB binding site located in the AD brain. Due to solubility-micellarization concerns that interferes with 3 H-PIB binding measurements, compounds were not tested at concentrations greater than 10 µM. These experiments were performed in duplicate (n = 1 independent experiment). The error bars represent standard deviations (SDEV).
To confirm the results obtained from the 3 H-PIB binding competition studies, we titrated the 1,3-and 1,4-chalcone-donepezil hybrids against 3 H-PIB bound to F 40 , F 42 , and ADPBC fibrils. The experiments were performed in duplicate for F 42 and ADPBC and as a single experiment for F 40 . Even though the single independent experiment may not be sufficient to determine the exact value of inhibition, we were able to observe a clear trend, which is sufficient at this stage of investigation. The data obtained by titration assays (Figure 2) were completely in accord with those displayed in Figure 1 for both the 1,3-and 1,4-chalcone-donepezil hybrids.

The Effect of Chalcone-Donepezil Hybrids on Aβ Assembly and Dissociation
The progression of AD neurodegenerative disease is characterized by the accumulation of Aβ plaques followed by the loss of neurons and tau pathology accumulation leading to brain atrophy and dementia. Even though Aβ fibrils are responsible for plaque formation, recent research efforts have pointed out the role of soluble Aβ oligomers in synaptic neurotoxicity [61]. Genetic evidence and autopsy observations of human cases of AD suggest that the buildup and pathogenicity of Aβ is fundamental to the progression of the disease. Since AD progression is tightly connected to Aβ aggregation, it is imperative to evaluate the effect of the 1,3-and 1,4-chalcone-donepezil hybrids 15a-15f and 16a-16f on Aβ oligomer assembly and dissociation. Among Aβ peptide assemblies, such as Aβ (1-40) (F 40 ) and Aβ (1-42) (F 42 ) fibrils, the Aβ (1-42) peptide is the most abundant and neuroand synaptotoxic [62][63][64][65].

The Effect of Chalcone-Donepezil Hybrids on Aβ Assembly and Dissociation
The progression of AD neurodegenerative disease is characterized by the accumulation of Aβ plaques followed by the loss of neurons and tau pathology accumulation leading to brain atrophy and dementia. Even though Aβ fibrils are responsible for plaque formation, recent research efforts have pointed out the role of soluble Aβ oligomers in synaptic neurotoxicity [61]. Genetic evidence and autopsy observations of human cases of AD suggest that the buildup and pathogenicity of Aβ is fundamental to the progression of the disease. Since AD progression is tightly connected to Aβ aggregation, it is imperative to evaluate the effect of the 1,3-and 1,4-chalcone-donepezil hybrids 15a-15f and 16a-16f on Aβ oligomer assembly and dissociation. Among Aβ peptide assemblies, such as Aβ(1-40) (F40) and Aβ(1-42) (F42) fibrils, the Aβ(1-42) peptide is the most abundant and neuro-and synaptotoxic [62][63][64][65].
We first examined the effect of the synthesized 1,3-and 1,4-chalcone-donepezil hybrids 15a-15f and 16a-16f on the assembly of N-biotinyl-Aβ(1-42) (bioAβ42) monomers into oligomers. We quantified the assembly of bioAβ42 into soluble oligomers by using an ELISA assay in which the soluble oligomers were captured on NeutrAvidin TM -coated ELISA plates and detected with streptavidinhorseradish peroxidase and colorimetric readout ( Figure 3). Among the hybrids, those with shorter We first examined the effect of the synthesized 1,3-and 1,4-chalcone-donepezil hybrids 15a-15f and 16a-16f on the assembly of N-biotinyl-Aβ (1-42) (bioAβ 42 ) monomers into oligomers. We quantified the assembly of bioAβ 42 into soluble oligomers by using an ELISA assay in which the soluble oligomers were captured on NeutrAvidin TM -coated ELISA plates and detected with streptavidin-horseradish peroxidase and colorimetric readout ( Figure 3). Among the hybrids, those with shorter linkers (2-8 carbons), 15a-15d and 16a-16d, demonstrated better inhibition of bioAβ 42 oligomer assembly than the parent chalcones 3 and 6 as well as the donepezil from which they were made, respectively. We found that hybrids with longer linkers (10 and 12 carbons), such as 15e, 15f, 16e, and 16f, did not prevent bioAβ 42 oligomerization. From these data, we could infer that attaching donepezil to chalcones is highly beneficial as it increases their ability to prevent assembly of bioAβ 42 oligomers.
Since we had shown in a previous publication that some chalcones could dissociate preformed bioAβ42 oligomers [58], we wanted to investigate if the covalent attachment of the various chalcones to donepezil, as in hybrids 15a-15f and 16a-16f, was responsible for the lack of ability of these hybrids to dissociate the oligomers. We wanted to determine if chalcones 3 or 17, or donepezil, or its analogue 20 alone could dissociate the preformed bioAβ42 oligomers ( Figure 7A). We also desired to see if combining chalcones 3 or 17 in a 1:1 mixture with donepezil or its analogue 20 would result in dissociation of bioAβ42 oligomers ( Figure 7B). We found that chalcones 3 and 17 alone, as well as donepezil and its analogue 20, did not dissociate the preformed oligomers. Neither did the various combinations tested. We also explored the addition of the ethylene glycol linker to donepezil, chalcone 3, and donepezil with chalcone 3 in 1:1 or 1:1:1 mixtures. We found that these combinations also did not allow for dissociation of preformed bioAβ42 oligomers. These data highlight that these specific chalcones 3 and 17 with a hydroxyl moiety at position 3, unlike the previously published chalcone with an hydroxyl at position 2 [58], do not dissociate preformed bioAβ42 oligomers. This was not completely surprising, as not all chalcones in our previous study were able to do so.

Molecular Modeling
In order to understand how our chalcone-donepezil hybrid molecules may bind simultaneously the AChE enzyme and an Aβ fibril, we built a model of a HsAChE-15a-Aβ fibril complex using a crystal structure of the HsAChE-donepezil complex (PDB ID: 4EY7 [66]) and a cryo-EM structure of an Aβ (residues 1-40) fibril (PDB ID: 6SHS [67]) ( Figure 8). The donepezil moiety of the hybrid was unambiguously defined by the bound donepezil in the former structure. In searching for a potential

Molecular Modeling
In order to understand how our chalcone-donepezil hybrid molecules may bind simultaneously the AChE enzyme and an Aβ fibril, we built a model of a HsAChE-15a-Aβ fibril complex using a crystal structure of the HsAChE-donepezil complex (PDB ID: 4EY7 [66]) and a cryo-EM structure of an Aβ (residues 1-40) fibril (PDB ID: 6SHS [67]) (Figure 8). The donepezil moiety of the hybrid was unambiguously defined by the bound donepezil in the former structure. In searching for a potential binding site for the chalcone-containing side of 15a in the Aβ fibril, we noticed that Glu22 and Asp23 formed a negatively-charged surface patch that could interact favorably with the terminal positively charged dimethyl amino group of 15a. The largely hydrophobic channel (lined with side chain of Phe19, Ala21, and Val24) would fit and favorably interact with the hydrophobic regions of the chalcone. The oxygen atom of the carbonyl group of chalcone may form hydrogen bonds with amide nitrogen atoms in this region of the Aβ, where β-sheets are highly distorted. Notably, the linker of 15a can span no more than two β-strand layers, as shown in Figure 8. In this model, the two β-strands are book-ended by the HsAChE on one side and the terminal dimethyl amino group on the other. These interactions would prevent propagation of the filament or even potentially cause a disassembly of a larger filament into at most two-strand-thick assemblies. It is obvious that in this model, a longer linker of the hybrid compound would allow a thicker filament to form; therefore, it would not be as potent in preventing filament propagation. binding site for the chalcone-containing side of 15a in the Aβ fibril, we noticed that Glu22 and Asp23 formed a negatively-charged surface patch that could interact favorably with the terminal positively charged dimethyl amino group of 15a. The largely hydrophobic channel (lined with side chain of Phe19, Ala21, and Val24) would fit and favorably interact with the hydrophobic regions of the chalcone. The oxygen atom of the carbonyl group of chalcone may form hydrogen bonds with amide nitrogen atoms in this region of the Aβ, where β-sheets are highly distorted. Notably, the linker of 15a can span no more than two β-strand layers, as shown in Figure 8. In this model, the two β-strands are book-ended by the HsAChE on one side and the terminal dimethyl amino group on the other. These interactions would prevent propagation of the filament or even potentially cause a disassembly of a larger filament into at most two-strand-thick assemblies. It is obvious that in this model, a longer linker of the hybrid compound would allow a thicker filament to form; therefore, it would not be as potent in preventing filament propagation.

Materials and Instrumentation
All chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA), Alfa Aesar (Ward Hill, MA, USA), and AK scientific (Union City, CA, USA), and used without further purification. Chemical reactions were monitored by thin layer chromatography (TLC) using Merck Silica gel 60 F254 plates. Visualization was achieved using UV light and a ceric molybdate stain (5 g (NH4)2Ce(NO3)6, 120 g (NH4)6Mo7O24·4H2O, 80 mL H2SO4, and 720 mL H2O). 1 H and 13 C NMR spectra were recorded at 400 and 100 MHz (or 125 MHz), respectively, on a Varian 400 MHz spectrometer (MR400) (or a Varian 500 MHz spectrometer; VNMRS 500), using the indicated deuterated solvents. Chemical shifts (δ) are given in parts per million (ppm). Coupling constants (J) are given in Hertz (Hz), and conventional abbreviations used for signal shape are as follows: s = singlet; d = doublet; t = triplet; m = multiplet; dd = doublet of doublets; ddd = doublet of doublet of doublets; br s = broad singlet; dt = doublet of triplets. Liquid chromatography-mass spectrometry (LCMS) was carried out using an Agilent 1200 series Quaternary LC system equipped with a diode array detector, and Eclipse XDB-C18 column

Materials and Instrumentation
All chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA), Alfa Aesar (Ward Hill, MA, USA), and AK scientific (Union City, CA, USA), and used without further purification. Chemical reactions were monitored by thin layer chromatography (TLC) using Merck Silica gel 60 F 254 plates. Visualization was achieved using UV light and a ceric molybdate stain (5 g (NH 4 ) 2 Ce(NO 3 ) 6 , 120 g (NH 4 ) 6 Mo 7 O 24 ·4H 2 O, 80 mL H 2 SO 4 , and 720 mL H 2 O). 1 H and 13 C NMR spectra were recorded at 400 and 100 MHz (or 125 MHz), respectively, on a Varian 400 MHz spectrometer (MR400) (or a Varian 500 MHz spectrometer; VNMRS 500), using the indicated deuterated solvents. Chemical shifts (δ) are given in parts per million (ppm). Coupling constants (J) are given in Hertz (Hz), and conventional abbreviations used for signal shape are as follows: s = singlet; d = doublet; t = triplet; m = multiplet; dd = doublet of doublets; ddd = doublet of doublet of doublets; br s = broad singlet; dt = doublet of triplets. Liquid chromatography-mass spectrometry (LCMS) was carried out using an Agilent 1200 series Quaternary LC system equipped with a diode array detector, and Eclipse XDB-C18 column (250 mm Plates were washed on a BioTek ELx50 plate washer (BioTek (Winooski, VT, USA)) and absorbance was read on a BioTek HT Synergy plate reader.

In Vitro Inhibition of EeAChE and EfBChE
The ChE inhibition assays were performed as previously described [29]. The 1,3-and 1,4-chalcone-donepezil hybrids (0.1 nM to 100 µM) were dissolved in sodium phosphate buffer ((100 µL), 0.1 M, pH 8.0). They were diluted 5-fold and either EeAChE or Ef BChE was added to the solution of the inhibitors (50 µL, containing 0.08 U/mL ChE (final concentration for both EeAChE and Ef BChE). The enzymes and inhibitors were incubated for 10 min followed by the addition of DTNB (50 µL, 0.25 mM final concentration) and acylthiocholine (acetylthiocholine for EeAChE and butyrylthiocholine for Ef BChE). The reactions were monitored at 412 nm by using a SpectraMax M5 plate reader (Molecular Devices, San Jose, CA) at 25 • C every 30 s for 10 min. Using the initial rates, the rate of no reaction was subtracted and normalized that value to the rate of no inhibitor. All assays were performed in triplicate. The data was plotted as a sigmoidal curve and IC 50 values were calculated using SigmaPlot 14.0 (Systat Software, San Jose, CA, USA). The IC 50 values for EeAChE and Ef BChE inhibitions are presented in Table 1 and the corresponding graphs are presented in Figures S66 and S67, respectively. In total, 250 µg of lyophilized NH 4 OH-treated Aβ (1-40) (cat # A-1157-02) and Aβ (1-42) (cat # A-1167-02) purchased from rPeptide (Watkinsville, GA, USA) were each solubilized in their glass vials with 250 µL of a buffer solution comprised of 20 mM NaP i , 145 mM NaCl, 0.02% w/v NaN 3 at pH 7.5 and incubated at 37 • C for 3 days, vortexing once per day. Prior to being transferred to screw-top polypropylene vials and being stored at −75 • C, these 1 mg/mL fibril suspensions were further diluted with an equal volume of the buffer solution. Their fibril content was determined in aliquots before and after centrifugation at 13,000× g by thioflavin fluorescence [73].

ADPBC Isolation
Human AD brain frontal cortical tissue (from the Sanders-Brown Center on Aging Alzheimer's Disease Center Brain Bank at the University of Kentucky) was processed as follows. The insoluble Pittsburgh Compound B-binding fraction (ADPBC = AD PIB-binding complex) was prepared from the frontal cortex by sequential differential centrifugation of the tissue homogenate followed by sodium dodecyl sulfate extraction [74]. After detergent extraction, the pellet was washed with 20 mM NaP i buffer with 145 mM NaCl at pH 7.5 and collected by centrifugation at 100,000× g and aliquots stored in screw-top polypropylene vials at −75 • C.

3 H-PIB Binding Assays
Competition of compounds for the binding of 3 H-PIB (VT 278, specific activity 70.2 Ci/mmol) purchased from Vitrax Radiochemicals (Placentia, CA) to Aβ  and Aβ (1-42) fibrils and AD brain extract (ADPBC) was determined by radioligand binding. 200 µL of 1.2 nM 3 H-PIB in 20 mM NaP i buffer with 145 mM NaCl at pH 7.5, and 5% v/v EtOH-containing compounds and solvent controls were added to 200 ng of Aβ  or Aβ  fibrils in 20 µL of 20 mM NaP i buffer with 145 mM NaCl at pH 7.5 in individual wells of a polypropylene 96-well plate (Costar 3355). For the ADPBC, 25 µL of 20 mM NaP i buffer with 145 mM NaCl at pH 7.5 containing the equivalent of 133 µg wet weight of original tissue were incubated with compounds and solvent controls. The plates were sealed and incubated for 3 h at 22 • C, without shaking, and transferred to a 96-well Millipore Multiscreen HTS Hi Flow FB (GF/B) filter plate, and filtered with a multi-well plate vacuum manifold (Millipore Corporation, Bedford, MA, USA). The filters were rapidly washed 4× with 200 µL of 20 mM NaP i buffer with 145 mM NaCl at pH 7.5, and 5% v/v EtOH, dried, and transferred to scintillation vials and Budget-Solve (Fisher) scintillation fluid was added. Specific binding of 3 H-PIB was determined by subtracting the non-specific binding determined in the presence of 1 µM BTA-1. These data are presented in Figure 1.

3 H-PIB Titration Assays
The titration of compounds with 3 H-PIB bound to Aβ (1-40) and Aβ (1-42) fibrils and AD brain extract (ADPBC) were performed to further determine the competition of our compounds for the 3 H-PIB binding site of the AD brain. The experiments were performed as described in the 3 H-PIB binding assays of Section 3.4.3 with the use of a concentration range of 10 to 0.03 µM for the compounds tested. These data are presented in Figure 2.

Preparation of ELISA Plates Coated with NeutrAvidin TM
Each well of ELISA plates (Costar 9018) were coated with 50 µL of 1 µg/mL NeutrAvidin TM in 10 mM NaP i buffer at pH 7.5 overnight at 4 • C. The next day, they were blocked with 200 µL of 20 mM NaP i buffer at pH 7.5 containing 0.145 M NaCl and 0.1% v/v Tween 20 for at least 2 h at rt. The plates were then stored in their blocked form at 4 • C prior to use for the oligomer assembly and dissociation assays.

Overview of bioAβ 42 Oligomer Assembly
The assembly of bioAβ 42 into soluble oligomers was quantified with an ELISA assay in which the soluble oligomers were captured with the NeutrAvidin TM coating the ELISA plates. The bound oligomers were detected with streptavidin-horseradish peroxidase (SA-HRP) [75]. The colorimetric or other signal produced was specific for oligomeric bioAβ 42 species because the biotin of the NeutrAvidin TM -captured monomeric bioAβ  was complexed with the NeutrAvidin TM and was unable to react with the SA-HRP. Using biotinylated reagents and biotin-binding proteins for capture and detection avoids the interaction of small molecules with antibodies and this selectivity is useful for screening compound libraries [75,76]. Below, please find the details of the three steps (Steps 1-3) used for the oligomer assembly assay.
Step 1: Preparation of monomeric bioAβ 42 . For a low background, it was essential to rigorously pretreat the bioAβ 42 peptide to convert (disaggregate) the multimeric species that form upon storage in the solid state, or even frozen in 1,1,1,3,3,3-hexafluoroisopropanol (HFIP) to monomers. The required amount of stock bioAβ 42 peptide (0.5 mg dissolved at 1 mg/mL in HFIP) was pipetted into a polypropylene microfuge tube, dried to a thin film under a gentle air or N 2 stream, and then disaggregated for 10 min at rt in neat trifluoroacetic acid (TFA). After removal of the TFA by drying with an air or N 2 stream, DMSO was added to produce a 2.3 µg/mL (50×) stock solution.
Step 2: Assembly of bioAβ 42 oligomers. 2 µL of disaggregated monomeric 50× bioAβ 42 peptide in DMSO (prepared in Step 1) was pipetted into the bottom of each well of a polypropylene 96-well plate. The assembly reaction was started by the addition of 100 µL of diluted compound containing up to 1% v/v DMSO in 20 mM NaP i buffer containing 0.145 M NaCl at pH 7.5. The 96-well plate was sealed and incubated at rt for 30 min without shaking. The bioAβ 42 oligomer assembly was stopped by adding 50 µL of 0.3% Tween 20 v/v in distilled H 2 O.
Step 3: Measurement of bioAβ 42 oligomers. To start measuring the bioAβ 42 oligomers, the blocking solution for the NeutrAvidin TM -coated ELISA plate was removed and 50 µL of bioAβ 42 oligomers (up to total of 20 nM bioAβ 42 in 20 mM NaP i buffer with 0.145 M NaCl and 0.1% v/v Tween 20 at pH 7.5) was added to each well. The 96-well plate was then sealed and incubated for 2 h while shaking at 150 rpm at rt. The plate was then washed 3× with 200 µL of a solution comprised of 20 mM Tris-HCl, 34 mM NaCl, 0.1% v/v Tween 20 at pH 7.5, and then 50 µL of SA-HRP (1:20,000) in 20 mM NaP i buffer with 0.145 M NaCl and 0.1% v/v Tween 20 at pH 7.5 was added to each well. The plate was sealed and incubated for 1 h while shaking at 150 rpm at rt. The plate was washed as previously described, and 100 µL/well of HRP substrate (i.e., 1 mM:0.07 mM/tetramethylbenzidine:H 2 O 2 ) in 0.2 M citrate buffer, pH 4.0 was added. The plate was incubated at rt for 5 to 10 min and the reaction was stopped by the addition of 100 µL/well of 1% v/v H 2 SO 4 . The absorbance of each well was read at 450 nm with a BioTek HT Synergy plate reader. These data are presented in Figure 3.

EC 50 Determination against bioAβ 42 Oligomer Assembly
The EC 50 values of hybrids 15a-15f and 16a-16f against bioAβ 42 oligomer assembly was determined as described in the overview of bioAβ 42 oligomer assembly in Section 3.5.2. A concentration range of 50 to 1.56 µM for the compounds was used in this study. These data are presented in Figure 4 and Table S1.

Overview of bioAβ 42 Oligomer Dissociation
Dissociation of preformed oligomers of bioAβ 42 was measured by pretreating biotinylated oligomers with solvent controls or compounds for a period of time, capturing them with NeutrAvidin TM on ELISA plates, and detecting the remaining bound oligomers with SA-HRP [77] as described above. Below, please find the details of the two steps (Steps 1 and 2) used for the oligomer dissociation assay.
Step 1: Preformed bioAβ 42 oligomers (sufficient for 400 wells). In total, 1 µg of bioAβ 42 peptide (1 mg/mL dissolved in HFIP) was pipetted into 20 µL of HFIP in a polypropylene microfuge tube, and dried to a thin film under a gentle air or N 2 stream. The film was dissolved in 250 µL of DMSO and vortexed. After 10 min at rt with intermittent vortexing, the DMSO-solubilized bioAβ 42 was added rapidly to 12.5 mL of 20 mM NaP i buffer with 0.145 M NaCl at pH 7.5 in a 17 × 100 mm polypropylene tube, sealed, and vortexed. After incubation at rt for 1 h with intermittent vortexing, 0.375 mL of 100 mM Tween 20 (10% v/v) in distilled H 2 O was added to make the final Tween 20 concentration 0.3% v/v. The preformed oligomers were mixed by inversion and aliquoted for storage at −75 • C. Note: Multiple freeze-thaw cycles should be avoided.
Step 2: Oligomer dissociation assay. Preformed bioAβ 42 oligomers (16.3 nM total bioAβ 42 ) in Tween 20 were added to diluted compound in 20 mM NaP i buffer with 0.145 M NaCl at pH 7.5 (final concentration of bioAβ 42 = 2.7 nM, ≤1% v/v DMSO) and incubated overnight (~18 h) with shaking. Then, 25 µL of the preformed bio42 oligomers adjusted to 0.6% v/v Tween 20 was pipetted into each well of a polypropylene 96-well plate. In total, 125 µL of diluted compound in 20 mM NaP i buffer with 0.145 M NaCl at pH 7.5 containing up to 1% v/v DMSO were added to the oligomers in the plates. The plates were sealed and shaken at 150 rpm overnight at rt. After 18 h of incubation, the plates were centrifuged at 1000× g for 10 min and a 100-µL aliquot was transferred from each well into a well of an NA-coated ELISA plate that had been blocked with Tween 20. The oligomer content of that aliquot was measured as described for the bioAβ 42 oligomer assembly assay. These data are presented in Figure 5 The effects of various chalcones, donepezil, donepezil analogue, and 1,3-chalcone-donepezil hybrids were tested for their ability to inhibit the bioAβ 42 oligomer assembly and dissociation either alone or in combination with each other. The experiments were performed as described in the biotinyl-Aβ (1-42) (bioAβ 42 ) oligomer assembly and dissociation assays section. 1:1 or 1:1:1 ratios of the compounds were used, and these data are presented in Figures 6 and 7.

Molecular Modeling
In the modeling of the HsAChE-15a-Aβ complex, we used the crystal structure of the HsAChE-donepezil complex (PDB ID: 4EY7 [66]) and the cryo-electron microscopy (cryo-EM) structure of an Aβ fibril (PDB ID: 6SHS [67]). The modeling was performed with Crystallographic Oriented Toolkit (Coot [78]) software. To generate the model, we superimposed the donepezil moiety of 15a with the bound donepezil in the HsAChE-donepezil crystal structure. The Aβ fibril was stripped down to two layers of Aβ and manually docked onto the chalcone side of 15a in an extended conformation while avoiding steric clashes and maximizing electrostatic and hydrophobic interactions between 15a and Aβ, assuming that the amino group of 15a would interact with the highly negatively-charged surface of Aβ defined by adjacent residues Glu22 and Asp23.

Conclusions
In summary, we synthesized 12 novel 1,3-and 1,4-chalcone-donepezil hybrids, 15a-15f and 16a-16f, and evaluated their ability to inhibit AD therapeutic targets. In general, most of these compounds were able to inhibit the actions of EeAChE and Ef BChE with low to sub-micromolar IC 50 values. In general, 1,4-chalcone-donepezil hybrids (16) were better at inhibiting EeAChE and Ef BChE than 1,3-chalcone-donepezil hybrids (15), with the exception of 15f against EeAChE and 15a and 15b against Ef BChE. The presence of shorter linkers between chalcones and donepezil seemed to have a positive impact on EeAChE and Ef BChE inhibition. In the displacement of 3 H-PIB from F 40 , F 42 , and ADPBC fibrils assays, we found that 1,3-and 1,4-chalcone-donepezil hybrids did not displace 3 H-PIB as effectively as chalcones 3 and 6, which pointed towards a different binding site for these hybrids compared to PIB. Compared to donepezil and parent 1,3-and 1,4-chalcones 3 and 6, compounds 15a, 15b, 15c, 15d, 16a, 16b, 16c, and 16d exhibited better inhibition of bioAβ 42 oligomer assembly. However, none of these 1,3-and 1,4-chalcone-donepezil hybrids were effective at dissociating preformed bioAβ 42 oligomers. Finally, we performed a combination study that convincingly showed that a covalent linkage between chalcones and donepezil is essential for the prevention of bioAβ 42 oligomerization. These studies demonstrate the promise of chalcone-donepezil hybrids as bifunctional molecules against two hallmarks of AD.
Supplementary Materials: The following are available online, the Supplementary Materials include 1 H and 13 C NMR spectra for all the molecules synthesized ( Figures S1-S65). The IC 50 curves for the inhibition of EeAChE ( Figure S66) and Ef BChE ( Figure S67) are also provided.