Design and Synthesis of Novel Hybrid 8-Hydroxy Quinoline-Indole Derivatives as Inhibitors of Aβ Self-Aggregation and Metal Chelation-Induced Aβ Aggregation

A series of novel hybrid 8-hydroxyquinoline-indole derivatives (7a–7e, 12a–12b and 18a–18h) were synthesized and screened for inhibitory activity against self-induced and metal-ion induced Aβ1–42 aggregation as potential treatments for Alzheimer’s disease (AD). In vitro studies identified the most inhibitory compounds against self-induced Aβ1–42 aggregation as 18c, 18d and 18f (EC50 = 1.72, 1.48 and 1.08 µM, respectively) compared to the known anti-amyloid drug, clioquinol (1, EC50 = 9.95 µM). The fluorescence of thioflavin T-stained amyloid formed by Aβ1–42 aggregation in the presence of Cu2+ or Zn2+ ions was also dramatically decreased by treatment with 18c, 18d and 18f. The most potent hybrid compound 18f afforded 82.3% and 88.3% inhibition, respectively, against Cu2+- induced and Zn2+- induced Aβ1–42 aggregation. Compounds 18c, 18d and 18f were shown to be effective in reducing protein aggregation in HEK-tau and SY5Y-APPSw cells. Molecular docking studies with the most active compounds performed against Aβ1–42 peptide indicated that the potent inhibitory activity of 18d and 18f were predicted to be due to hydrogen bonding interactions, π–π stacking interactions and π–cation interactions with Aβ1–42, which may inhibit both self-aggregation as well as metal ion binding to Aβ1–42 to favor the inhibition of Aβ1–42 aggregation.


Introduction
Alzheimer's disease (AD) is the most debilitating age-associated neurodegenerative disorder leading to dementia, affecting millions of elderly people, and the number of patients is expected to reach 130 million worldwide by 2050 [1,2]. AD remains incurable due to the low efficacy and the very limited number of available drugs to treat this neurodegenerative disease. Salient features of AD are the accumulation of amyloid-β (Aβ) plaques and tangles containing hyperphosphorylated tau protein. These proteins misfold and aggregate in the brains of affected individuals as amyloid Taking into account neuronal loss, deficits in adult neurogenesis, and the formation of plaques and tangles associated with AD, the present work is focused on the identification of pharmacophores containing the 8-hydroxyquinoline moiety conjugated to an indole scaffold, in order to generate a series of hybrid 8-hydroxyquinoline indole esters and amides without (7a-7e and 12a-12b) and with a piperazine bridging moiety (18a-18h). Fifteen novel hybrid 8-hydroxyquinoline-indole analogs of this type were prepared and investigated for their inhibition of Aβ1-42 self-aggregation and metal ioninduced aggregation. Compounds which showed inhibitory activity against self-induced Aβ1-42 aggregation were also evaluated against HEK-tau and SY5Y-APPSw cells to determine their effects on cellular protein aggregation. For the synthesis of hybrid 8-hydroxyquinoline indoles (7a-7e), we initially prepared tert-butyl (2-(hydroxymethyl)quinolin-8-yl) carbonates 4a and 4b utilizing literature procedures [35], and carried out EDC coupling of each compound with a variety of indole-2-carboxylic acids (5a-5c) to afford Boc-protected ester intermediates 6a-6e. Boc-deprotection of intermediates 6a-6e was carried out with trifluoroacetic acid in dichloromethane at room temperature to afford 7a-7e (Scheme 1).  Taking into account neuronal loss, deficits in adult neurogenesis, and the formation of plaques and tangles associated with AD, the present work is focused on the identification of pharmacophores containing the 8-hydroxyquinoline moiety conjugated to an indole scaffold, in order to generate a series of hybrid 8-hydroxyquinoline indole esters and amides without (7a-7e and 12a-12b) and with a piperazine bridging moiety (18a-18h). Fifteen novel hybrid 8-hydroxyquinoline-indole analogs of this type were prepared and investigated for their inhibition of Aβ1-42 self-aggregation and metal ioninduced aggregation. Compounds which showed inhibitory activity against self-induced Aβ1-42 aggregation were also evaluated against HEK-tau and SY5Y-APPSw cells to determine their effects on cellular protein aggregation. For the synthesis of hybrid 8-hydroxyquinoline indoles (7a-7e), we initially prepared tert-butyl (2-(hydroxymethyl)quinolin-8-yl) carbonates 4a and 4b utilizing literature procedures [35], and carried out EDC coupling of each compound with a variety of indole-2-carboxylic acids (5a-5c) to afford Boc-protected ester intermediates 6a-6e. Boc-deprotection of intermediates 6a-6e was carried out with trifluoroacetic acid in dichloromethane at room temperature to afford 7a-7e (Scheme 1). Taking into account neuronal loss, deficits in adult neurogenesis, and the formation of plaques and tangles associated with AD, the present work is focused on the identification of pharmacophores containing the 8-hydroxyquinoline moiety conjugated to an indole scaffold, in order to generate a series of hybrid 8-hydroxyquinoline indole esters and amides without (7a-7e and 12a-12b) and with a piperazine bridging moiety (18a-18h). Fifteen novel hybrid 8-hydroxyquinoline-indole analogs of this type were prepared and investigated for their inhibition of Aβ 1-42 self-aggregation and metal ion-induced aggregation. Compounds which showed inhibitory activity against self-induced Aβ 1-42 aggregation were also evaluated against HEK-tau and SY5Y-APP Sw cells to determine their effects on cellular protein aggregation.
The product was then reacted with sodium azide, at reflux temperature in acetone for 6 h, to afford the azido intermediate 9. Compound 9 was then reduced to the amino compound 10 by treatment with PPh3 in water at 60 °C for 12 h (Scheme 2).  Finally, EDC coupling of 10 with indole-2-carboxylic acids 5a or 5b afforded the Boc-protected amide intermediates 11a and 11b. Boc-deprotection of 11a and 11b was carried out with trifluoroacetic acid in dichloromethane at room temperature to afford 12a and 12b, respectively (Scheme 3).
Based on the above results from self-induced Aβ 1-42 aggregation and metal-induced Aβ 1-42 aggregation studies, the most potent compounds 18f was selected for further investigation into its ability to chelate metal ions such as Cu 2+ , Zn 2+ , and Fe 2+ using UV-vis spectroscopy at wavelengths ranging from 200 to 500 nm. Compound 18f (50 µM) was treated with 50 µM concentrations of CuSO 4 , ZnCl 2 , or FeSO 4 for 30 min in HEPES buffer at room temperature, and displayed maximum absorption wavelength shifts from 252 to 274, 269 and 255 nm in the presence of Cu 2+ , Zn 2+ , and Fe 2+ , respectively ( Figure 5). These wavelength shifts are indicative of the formation of 18f-Cu 2+ , 18f-Zn 2+ , and 18f-Fe +2 and show that Zn 2+ and Cu 2+ undergo significant chelation with compound 18f. The spectrum of 18f-Fe 2+ showed only a moderate increase in wavelength. Furthermore, the stoichiometry of formation of metal complexes 18f-Cu 2+ and 18f-Zn 2+ was also determined by using Job's method ( Figure 6) [36]. The isosbestic points shown in Figure 6A,B at 0.6 and 0.5 imply a stoichiometry of 1.5:1 for 18f-Cu 2+ and 1:1 for 18f-Zn 2+ .  Copper and zinc dysregulation in the brain is considered to arise as a consequence of age-associated neurodegenerative disease such as Alzheimer's disease. Zinc has also been shown to play an important role in pre-and post-synaptic responses [37]. Our drug design approach is to target dysregulation of metal homeostasis and protein aggregation. Dysregulation of metal ions is implicated in the elevated formation of reactive oxygen species and in susceptibility to oxidative stress. These metal ions were shown to interact with protein aggregation seed proteins such as tau and Aβ 1-42 , increasing misfolding and promoting protein aggregation. The 8-hydroxyquinoline compound, PBT2, has been shown to influence trans-synaptic effects of copper and zinc ions in a cell culture model and has been considered as a treatment for AD [38]. PBT2 (2) was also shown to induce amyloid degradation by inhibiting glycogen synthase kinase-3 activity via phosphorylation [38].  In our study, we have shown that restoring metal homeostasis by treatment with hybrid 8hydroxyquinoline-indole analogs reduces tau-and amyloid-mediated protein aggregation, and their associated neurotoxicity. Analogs 7a-7d, 18c-18f, and 18h were evaluated for their ability to reduce protein aggregation in HEK-tau cells. HEK-tau cells were exposed to the above Aβ1-42 aggregation inhibitors and parent drug clioquinol (1) at 1 µM for 48 h at 37 °C. Cells were assessed for amyloid deposits by staining with 0.1% w/v thioflavin T, counterstaining nuclei with DAPI, and mean ThT fluorescence signal per nucleus was calculated over multiple fields [39]. All the analogs reduced fluorescence intensity (indicative of protein aggregation) by 26-62% (P < 0.001 by 2-tailed t-test). Among these analogs, compounds 7b-7d, 18c, and 18f were the most effective as protein aggregation inhibitors in HEK-tau cells, and were more potent than clioquinol (1) (Figure 7A,B). In our study, we have shown that restoring metal homeostasis by treatment with hybrid 8-hydroxyquinoline-indole analogs reduces tau-and amyloid-mediated protein aggregation, and their associated neurotoxicity . Analogs 7a-7d, 18c-18f, and 18h were evaluated for their ability to reduce protein aggregation in HEK-tau cells. HEK-tau cells were exposed to the above Aβ 1-42 aggregation inhibitors and parent drug clioquinol (1) at 1 µM for 48 h at 37 • C. Cells were assessed for amyloid deposits by staining with 0.1% w/v thioflavin T, counterstaining nuclei with DAPI, and mean ThT fluorescence signal per nucleus was calculated over multiple fields [39]. All the analogs reduced fluorescence intensity (indicative of protein aggregation) by 26-62% (P < 0.001 by 2-tailed t-test). Among these analogs, compounds 7b-7d, 18c, and 18f were the most effective as protein aggregation inhibitors in HEK-tau cells, and were more potent than clioquinol (1) (Figure 7A,B). In another study compounds 7a-7d, 18c-18f, and 18h were also evaluated for their ability to reduce protein aggregation in SY5Y-APPSw cells. SY5Y-APPSw cells were exposed to the above protein aggregation inhibitors and parent compound 1 at 1 µM for 48 h at 37 °C. Cells were assessed for amyloid deposits by staining with 0.1% w/v ThT, counterstaining nuclei with DAPI, and mean ThT fluorescence signal per nucleus calculated over multiple fields. All the analogs reduced fluorescence intensity (indicative of protein aggregation) by 22-67% (P < 0.001 by 2-tailed t-test). Analogs 7a and In another study compounds 7a-7d, 18c-18f, and 18h were also evaluated for their ability to reduce protein aggregation in SY5Y-APP Sw cells. SY5Y-APP Sw cells were exposed to the above protein aggregation inhibitors and parent compound 1 at 1 µM for 48 h at 37 • C. Cells were assessed for amyloid deposits by staining with 0.1% w/v ThT, counterstaining nuclei with DAPI, and mean ThT fluorescence signal per nucleus calculated over multiple fields. All the analogs reduced fluorescence intensity (indicative of protein aggregation) by 22-67% (P < 0.001 by 2-tailed t-test). Analogs 7a and 18d were the most effective in SY5Y-APP Sw cells and were more potent than the parent compound, clioquinol (1) (Figure 7C,D). Thioflavin-T fluorescence assays for in vitro amyloid aggregation and cell culture models of amyloid aggregation overlap but are not identical. This is due to the fact that thioflavin in cell-culture stains not just amyloid aggregates but also other amyloid-like fibrillar structures. Another important difference is that in vitro aggregation is assessed after 24 h, whereas aggregation in cell culture is assessed after 48 h.
The binding interactions of the active compounds 18d and 18f were also investigated utilizing a crystal structure of the target Aβ 1-42 peptide (PDB code: 1IYT) [40] employing Schrodinger Maestro 11.4 software (Figure 8) [41]. For analogue 18d, the quinoline ester carbonyl group of this hybrid molecule is involved in a hydrogen bonding interaction with Lys16 at a distance of 3.12 A o units and the 8-hydroxyquinoline moiety is involved in a π-π stacking interaction with residue Phe20; this is similar to that observed with the deoxyvasicinone-donepezil hybrid molecule binding region on Aβ 1-42 reported in the literature [41]. In addition, the quinolone and indole ring systems of compound 18d are involved in π-cation interactions with Lys16. With the potent analog 18f, the hydroxyl group of the 8-hydroxyquinoline moiety is involved in a hydrogen bonding interaction with the His13 residue at 2.17 A o units, and the 8-hydroxyquinoline moiety of 18f is also involved in a π-π stacking interaction with Phe20, similar to the binding mode of compound 18d. In addition, the quinoline ring system of 18f is involved in a π-cation interaction with Lys16. These docking results indicate that the effective inhibition of both self-induced and metal ion-induced Aβ 1-42 aggregation by 18d and 18f may be due to hydrogen bonding interactions, π-π stacking interactions, and π-cation interactions of these molecules. Of particular interest is the interaction the most potent compound 18f with the His13 residue on the Aβ 1-42 peptide. His13 has been implicated in the binding of metal ions with Aβ 1-42 (17)(18)(19)(20)(21); thus, 18f may compete with metal ions for this binding site on Aβ 1-42 , thereby inhibiting metal ion induction of Aβ 1-42 aggregation.  In summary, a series of hybrid 8-hydroxyquinoline-indole derivatives 7a-7e, 12a-12b, and 18a-18h have been synthesized and evaluated as inhibitors of Aβ1-42 self-aggregation and metal chelationinduced aggregation, and have also been evaluated against HEK-tau and SY5Y-APPSw cells to determine their effect on cellular protein aggregation. In vitro studies showed that hybrid 8hydroxyquinoline-indole analogs lacking the piperazine bridging unit (i.e., compounds 7a-7e and 12a-12b) exhibited poor-to-moderate inhibitory activities against Aβ1-42 peptide self-aggregation. However, the majority of the hybrid 8-hydroxyquinoline-indole compounds incorporating the piperazine bridge unit (i. e., 18c-18f, and 18h) were potent inhibitors against Aβ1-42 self-aggregation. Analogue 18f was the most potent inhibitor of Aβ1-42 aggregation in both self-induced and metal chelation-induced assays, and in protein aggregation assays in HEK-tau cells. Compound 18f afforded 82.3% and 88.3% inhibition against Cu 2+ -induced and Zn 2+ -induced Aβ1-42 aggregation assays, and was also shown to be a chelator of Cu 2+ , Zn 2+ , and Fe 2+ ions in physiological buffer solutions. Molecular docking studies with 18f showed that this molecule has the potential to interact with the His13 residue on Aβ1-42 peptide, which suggests that its potent anti-aggregation properties In summary, a series of hybrid 8-hydroxyquinoline-indole derivatives 7a-7e, 12a-12b, and 18a-18h have been synthesized and evaluated as inhibitors of Aβ 1-42 self-aggregation and metal chelation-induced aggregation, and have also been evaluated against HEK-tau and SY5Y-APP Sw cells to determine their effect on cellular protein aggregation. In vitro studies showed that hybrid 8-hydroxyquinoline-indole analogs lacking the piperazine bridging unit (i.e., compounds 7a-7e and 12a-12b) exhibited poor-to-moderate inhibitory activities against Aβ 1-42 peptide self-aggregation. However, the majority of the hybrid 8-hydroxyquinoline-indole compounds incorporating the piperazine bridge unit (i. e., 18c-18f, and 18h) were potent inhibitors against Aβ 1-42 self-aggregation .
Analogue 18f was the most potent inhibitor of Aβ 1-42 aggregation in both self-induced and metal chelation-induced assays, and in protein aggregation assays in HEK-tau cells. Compound 18f afforded 82.3% and 88.3% inhibition against Cu 2+ -induced and Zn 2+ -induced Aβ 1-42 aggregation assays, and was also shown to be a chelator of Cu 2+ , Zn 2+ , and Fe 2+ ions in physiological buffer solutions. Molecular docking studies with 18f showed that this molecule has the potential to interact with the His13 residue on Aβ 1-42 peptide, which suggests that its potent anti-aggregation properties may also be due in part to its ability to competitively inhibit the binding of metal ions that interact with this site to induce Aβ 1-42 aggregation. Thus, the present study has identified 18f, an analog that incorporates a piperazine bridging group between a 5-methoxyindole moiety and a 5,7-dichloro-8-hydroxyquinoline moiety, as the most potent Aβ 1-42 aggregation inhibitor; this analog exhibited a 10-fold improvement in inhibitory potency over clioquinol (1) as an inhibitor of Aβ 1-42 self-aggregation. We consider 18f to be a lead compound worthy of further structural optimization and preclinical development as a treatment for AD.

Chemistry
All reagents, solvents, and chemicals utilized in the synthesis of the hybrid 8-hydroxyquinoline-indole analogs were purchased from Oakwood Chemicals Fisher Scientific and Adooq Bioscience. 1 H and 13 C-NMR spectra were recorded on a Varian 400 MHz spectrometer equipped with a Linux workstation running on vNMRj software. Spectral analyses were carried out in CDCl 3 or DMSO-d 6 for both 1 H and 13 C spectra. Chemical shifts were measured in δ parts per million (ppm) and coupling constants (J) were measured in hertz (Hz). High-resolution mass spectra (HR-MS) were recorded on an Agilent 6545 ESI/APCI TOF MS. Thin-layer chromatography (TLC) was carried out on pre-coated silica gel glass plates (F254 Merck).

Methodology for the In Vitro Self-Induced Aβ 1-42 Aggregation Assay
The thioflavin T (ThT)-fluorescence assay was used to measure the inhibition of Aβ aggregation [42]. Aβ 1-42 (Adooq Bioscience, CA, USA) was pretreated with 1 mL of hexafluoroisopropanol (HFIP) to afford a stock solution, which was aliquoted into small samples. The solvent was evaporated at room temperature, and samples were stored at −80 • C. For Aβ 1-42 aggregation inhibition experiments, phosphate buffer (pH 7.4) was added to the Aβ stock solution to afford a 50 µM concentration before use. A mixture of the Aβ 1-42 peptide (10 µL, 25 µM final concentration) with or without the test compound (10 µL; 0.3, 1.0, 3, 9, and 27 µM) was incubated at 37 • C for 48 h. Blanks using phosphate buffer (pH 7.4) instead of Aβ, with or without test compound, were also assessed. Then 50 mM glycine-NaOH buffer (pH 8.0) containing ThT (5 µM) was added to 20 µL of the sample to afford a final volume of 200 µL. The fluorescence intensities were recorded (excitation, 450 nm; emission, 485 nm). The fluorescence intensities were background corrected to the no enzyme control and 100% of intensity reflected no inhibition of aggregation. To obtain the normalized amyloid aggregation inhibition relative to the control group, the amyloid aggregation/cell values were further divided by the total amyloid fluorescence of the control group. Using GraphPad Prism, the amyloid aggregation values (relative to control group) were entered along the corresponding concentrations for the triplicates and then a log transformation was obtained. Using the dose-response simulation module in GraphPad Prism, nonlinear regression fit (curve) was then generated for each small molecule with various concentrations used. Curves were then compared to determine if they were statistically different, by performing the best-fit values of selected unshared parameters between data sets and extra sum-of-squares F test, allowing for selection of the simpler model unless the P value was less than 0.05. The logEC 50 output parameter was then selected to obtain the EC 50 for each of the dose-response curves for the small molecules.

Methodology for the In Vitro Metal-Induced Aβ 1-42 Aggregation Assay
For the inhibition of copper-and zinc-mediated Aβ 1-42 aggregation, 20 µM HEPES (pH 6.6) in 150 µM NaCl was added to the Aβ 1-42 stock solution to afford a 25 µM solution. Mixtures of the peptide (10 µL, 25 µM final concentration) with or without Cu 2+ or Zn 2+ (10 µL, 25 µM final concentration) and the test compound (10 µL, 50 µM final concentration) were incubated at 37 • C for 24 h. Then 50 mM glycine-NaOH buffer (pH 8.0) containing ThT (5 µM) was added to 20 µL of the sample and diluted to afford a final volume of 200 µL. Fluorescence intensities of the solutions were recorded (excitation, 450 nm; emission, 485 nm). The percentage of aggregation was calculated by the expression (1 − IFi/IFc) × 100, in which IFi and IFc are the fluorescence intensities obtained for Aβ in the presence and absence of inhibitors, respectively.

Methodology for the Metal-Chelation Study
The chelating studies were performed with a UV-vis spectrophotometer. The absorption spectra of test compound (50 µM, final concentration) alone or in the presence of CuSO 4 , FeSO 4 , or ZnCl 2 (50 µM, final concentration) in buffer (20 mM HEPES, 150 mM NaCl, pH 7.4) were incubated for 30 min and then recorded at room temperature, respectively. For the stoichiometry of the test compound-Cu 2+ complex or the test compound-Zn 2+ complex, a fixed amount of test compound (50 µM) was mixed with increasing amounts of copper ion or Zn ion (0-100 µM), and the UV-vis difference spectra were analyzed to determine the ratio of ligand/metal ion in the complex.

Methodology for In Vitro HEK-Tau and SY5Y-APP Sw Cell Aggregation Assays
Human Embryonic Kidney cells that overexpress tau protein (HEK-tau) cells or SY5Y neuroblastoma cells that express a familial-AD mutant of amyloid precursor protein (SY5Y-APP Sw cells) were each seeded in 96-well plates at 8000 cells per well. After 24 h, cells were supplemented with test compound at 1 µM after diluting with medium. Vehicle only control was also included for each experiment. The cells were cultured for an additional 48 h at 37 • C. Both the cell lines were able to reach approximately 80% confluency in 48 h. Cells were assessed for amyloid deposits by staining with 0.1% w/v ThT, counterstaining nuclei with DAPI, and calculating mean ThT fluorescence signal per nucleus over multiple fields.

Methodology for Molecular Docking
Molecular docking was performed using Schrodinger software Maestro 11.4 suite. The ligands were prepared using the LigPrep module. The protein crystal structure PDB ID: 1IYT was downloaded from the RCSB Protein Data Bank (https://www.rcsb.org/). The downloaded protein crystal structure was prepared in Protein Preparation Wizard wherein the missing hydrogens were added and bond orders assigned. A grid for docking the ligand was generated with the centroid of the co-crystallized ligand from PDB ID: 1IYT as the center of the grid with no constraints. The prepared ligands 18d and 18f were docked into this grid using Glide module. Molecular docking was performed at XP precision and results were analyzed in the Maestro visualizer.