A Unique GSK-3β inhibitor B10 Has a Direct Effect on Aβ, Targets Tau and Metal Dyshomeostasis, and Promotes Neuronal Neurite Outgrowth

Due to the complicated pathogenesis of Alzheimer’s disease (AD), the development of multitargeted agents to simultaneously interfere with multiple pathological processes of AD is a potential choice. Glycogen synthase kinase-3β (GSK-3β) plays a vital role in the AD pathological process. In this study, we discovered a novel 1H-pyrrolo[2,3-b]pyridine derivative B10 as a GSK-3β inhibitor that features with a quinolin-8-ol moiety to target the metal dyshomeostasis of AD. B10 potently inhibited GSK-3β with an IC50 of 66 ± 2.5 nM. At the concentration of 20 μM, B10 increased β-catenin abundance (β-catenin/GAPDH: 0.83 ± 0.086 vs. 0.30 ± 0.016), phosphorylated GSK-3β at Ser9 (p-GSK-3β/GAPDH: 0.53 ± 0.045 vs. 0.35 ± 0.012), and decreased the phosphorylated tau level (p-tau/GAPDH: 0.33 ± 0.065 vs. 0.83 ± 0.061) in SH-SY5Y cells. Unlike other GSK-3β inhibitors, B10 had a direct effect on Aβ by inhibiting Aβ1-42 aggregation and promoting the Aβ1-42 aggregate disassociation. It selectively chelated with Cu2+, Zn2+, Fe3+, and Al3+, and targeted AD metal dyshomeostasis. Moreover, B10 effectively increased the mRNA expression of the recognized neurogenesis markers, GAP43, N-myc, and MAP-2, and promoted the differentiated neuronal neurite outgrowth, possibly through the GSK-3β and β-catenin signal pathways. Therefore, B10 is a potent and unique GSK-3β inhibitor that has a direct on Aβ and serves as a multifunctional anti-AD agent for further investigations.


Chemicals, Reagents, Cell Lines, and Antibodies
All the chemicals and reagents for the synthesis of B10 were commercially available or prepared according to the cited literature. A Bruker-600 NMR spectrometer (Brucker Co., Ltd., Zurich, Switzerland) was used to measure the 1 H and 13 C NMR spectra. High-resolution mass spectra (HRMS) were determined on a 6520 QTOF instrument (Agilent Technologies Inc., Santa Clara, CA, USA) by electrospray ionization (ESI). Melting points were measured on an X-6 micromelting point apparatus (Beijing Tech. Co., Ltd., Beijing, China). Column chromatography was performed on silica gel (200-300 mesh). In cell assay, compound B10 was dissolved in 100% dimethylsulfoxide (DMSO, Sigma, Shanghai, China) to obtain a 2 mM stock solution, the stock solutions were diluted in culture medium at various concentrations before each treatment, and the final DMSO concentration did not exceed 0.05% (v/v). SH-SY5Y cell line was provided by JiangShu KeyGen Biotech Co., Ltd. GSK3β was purchased from CarnaBio (Natick, MA, USA). Aβ 1-42 (ChinaPeptides Co., Ltd., Shanghai, China) was dissolved in 100% DMSO to form a 2 mM stock solution immediately before use. p-GSK3β (Ser9) antibody was obtained from Abcam China (Shanghai). The β-catenin and p-tau (Ser396) were purchased from JiangShu KeyGen Biotech Co., Ltd. All other chemicals or reagents were of analytical grade and were obtained from Sigma Chemical Co. (Shanghai, China).

Glycogen Synthase Kinase-3β (GSK-3β) Kinase Assay
The inhibitory activity of B10 against GSK-3β was determined by the caliper mobility shift assay and followed the manufacturer protocol, using staurosporine as a positive control. Staurosporine or B10 was tested from 1 µM or 5 µM, 3-fold dilution, in IC 50 determination. The kinase reaction was done in 96-well plate (Corning, Los Altos, MA, USA). Each well was loaded with compound and GSK-3β. The mixture was incubated at room temperature for 10 min. The reaction was started by the addition of peptide FAM-P15 (GL Biochem, Shanghai, China) and ATP (Sigma, Shanghai, China) prepared in reaction buffer. After incubation at 28 • C for 1 h, a stop buffer (25 µL) was added. The stopped reaction was analyzed on a LabChip EZ Reader (PerkinElmer, Shanghai, China) to give the conversion data at each concentration through the direct detection of both substrate and product via Laser-Induced Fluorescence (LIF) at 492 nm. The IC 50 values were then calculated from dose-response curves using XLfit (curve fitting software for Excel).

Metal Chelation
Compound B10 was dissolved in 100% methanol to form a 20 µM solution. The salt (NaCl, KCl, CaCl 2 , MgCl 2 , FeSO 4 , ZnCl 2 , CuCl 2 , and AlCl 3 ) was dissolved in 100% methanol to give a 40 µM metal ion solution. A mixture of the B10 solution (20 µM, 2 mL) with the metal ion solution (40 µM, 2 mL) or with 100% methanol (2 mL) was incubated at 25 • C for 30 min. The UV spectra of B10 alone or in the presence of each metal ion were recorded by TU-1901 ultraviolet-visible spectrophotometer (Beijing PurSee General Co, Ltd, Beijing, China) at the wavelength ranging from 200 to 500 nm, respectively.  Aggregation and disaggregation of Aβ 1−42 during the ThT assays were confirmed by transmission electron microscopy (TEM). Briefly, aliquots (10 µL) of the samples were taken from the wells and placed on a carbon-coated copper grid for 20 min. Each grid was stained with aqueous phosphotungstic acid (2%, 10 µL) for 1 min. After draining the excess staining solution, the specimen was transferred for imaging by TEM (Hitachi HT7700, Tokyo, Japan).

Neuronal Neurite Outgrowth Assay and Quantitative Real-Time Reverse Transcription-PCR (RT-PCR)
SH-SY5Y cells were cultivated in DMEM containing 10% FBS, 100 IU/L penicillin, and 10 µg/mL streptomycin and seeded in 96-well plates (Corning, Los Altos, MA, USA) at a density of 5 × 10 3 in 100 µL medium per well and kept in 5% CO 2 atmosphere at 37 • C for 24 h. After that, compound (RA or B10) diluted in medium (100 µL) was added and cultivated for 72 h. The morphology of neurite outgrowth was examined under an inverted microscope (2 × 100; Olympus, Tokyo, Japan).

Molecular Docking
The protein structures of GSK-3β (PDB ID: 5F95) and Aβ 1−42 monomer (PDB codes: 1IYT) were obtained from the RCSB Protein Date Bank. The protein structure was prepared with the SYBYL-X suite (version 2.1.1, Tripos). For GSK-3β, the ligand was first extracted from the protein, and then the hydrogen atoms and charges were added. The extracted ligand was used as a standard to generate the protomol. For Aβ 1−42 , hydrogen atoms were added to the crystal and charges were added to biopolymer by AMBER7 FF99 force field. The protomol was generated according to residues 13-27 (HHQKLVFFAEDVGSN). The 3D structures of B10 were prepared by the Sketch module of Sybyl Cells 2020, 9, 649 6 of 15 followed by energy minimization using the Tripos force field. The Surflex-dock module was used for the simulations, and the related parameters implied in the program were kept at default.

Statistics
Qualitative data including the immunoblots and images are representatives of at least three independent experiments and expressed as means ± SD. Statistical differences between two groups were determined by the two-tailed unpaired or paired Student t-test. p < 0.05 is considered significantly different.

Design and Synthesis of B10
Metal dyshomeostasis plays important roles in AD pathogenesis by preceding or inducing NFTs, senile plaques, and reactive oxygen species (ROS) [14][15][16][17]. Cu 2+ , Zn 2+ , and Fe 2+ are known to excessively exist in the senile plaques [17,40]. As one of the main cationic elements in plaque formation, copper ion binds to Aβ, promoting its accumulation and inducing ROS generation and oxidative stress [41,42]. The generated ROS may in turn produce modified Aβ species that favor Aβ aggregation and resist disaggregation [43,44]. 8-Hydroxyquinoline is a bidentate metal chelator and one of the most-used metal-chelating scaffolds in the design of multitarget AD agents [41,45]. To target the multifacets of AD, we designed novel pyrrolo [2,3-b]pyridinyl-based GSK-3β inhibitors incorporating 8-hydroxyquinoline motif to also target AD metal dyshomeostasis. The synthesis of B10 is shown in Figure 1. Qualitative data including the immunoblots and images are representatives of at least three independent experiments and expressed as means ± SD. Statistical differences between two groups were determined by the two-tailed unpaired or paired Student t-test. p < 0.05 is considered significantly different.

B10 Is a Potent GSK-3β In Vitro Inhibitor
Staurosporine is a prototypical ATP-competitive kinase inhibitor that potently inhibits GSK-3β with a reported IC50 value of 15 nM [48]. In the caliper mobility shift assay, staurosporine was used as a positive control with a determined IC50 of 16.5 ± 1.2 nM (n = 3). B10 was found to be a potent GSK-3β inhibitor with an IC50 of 66 ± 2.5 nM (Figure 2A).

B10 Is a Potent GSK-3β In Vitro Inhibitor
Staurosporine is a prototypical ATP-competitive kinase inhibitor that potently inhibits GSK-3β with a reported IC 50 value of 15 nM [48]. In the caliper mobility shift assay, staurosporine was used as a positive control with a determined IC 50 of 16.5 ± 1.2 nM (n = 3). B10 was found to be a potent GSK-3β inhibitor with an IC 50 of 66 ± 2.5 nM (Figure 2A). formed two hydrogen bonds with VAL135 and ASP133 in the hinge region; the 8-hydroxyquinoline scaffold occupied one hydrophobic pocket and made crucial hydrogen bonding with LYS85 through its 8-hydroxyl group; the pyridin-4-amine portion fell into another hydrophobic pocket and formed hydrogen bond interaction with ARG141 through the pyridine nitrogen atom. All these hydrophobic interactions and hydrogen bonds may contribute to its high potency against GSK-3β.

B10 Regulates GSK-3β and β-catenin Signal Pathways and Inhibits Tau Phosphorylation in Human Neuroblastoma SH-SY5Y Cells
Since B10 showed high potency in the inhibition of GSK-3β at the enzymatic level, we further investigated its effect on SH-SY5Y cells by Western blot assays. LiCl is a non-ATP competitive inhibitor of GSK-3β. At the concentration of 5 mM, LiCl greatly increased the levels of GSK-3β phosphorylated at Ser9 (p-GSK-3β/GAPDH: 0.62 ± 0.057 vs. 0.35 ± 0.012), an inactive form of GSK-3β ( Figure 3A). Treatment with B10 at 10 μM and 20 μM dose-dependently increased the phospho-GSK-3β level in comparison with the control group, and the p-GSK-3β/GAPDH ratio was 0.44 ± 0.037 and 0.53 ± 0.045, respectively.  To explore the possible interaction mode of the B10 with GSK-3β, molecular docking studies were performed on SYBYL based on the reported GSK-3β structure (PDB ID: 5F95) [49]. As shown in Figure 2B, B10 fitted well into the ATP binding pocket of GSK-3β: the pyrrolo[2,3-b]pyridinyl moiety formed two hydrogen bonds with VAL135 and ASP133 in the hinge region; the 8-hydroxyquinoline scaffold occupied one hydrophobic pocket and made crucial hydrogen bonding with LYS85 through its 8-hydroxyl group; the pyridin-4-amine portion fell into another hydrophobic pocket and formed hydrogen bond interaction with ARG141 through the pyridine nitrogen atom. All these hydrophobic interactions and hydrogen bonds may contribute to its high potency against GSK-3β.

B10 Regulates GSK-3β and β-catenin Signal Pathways and Inhibits Tau Phosphorylation in Human Neuroblastoma SH-SY5Y Cells
Since B10 showed high potency in the inhibition of GSK-3β at the enzymatic level, we further investigated its effect on SH-SY5Y cells by Western blot assays. LiCl is a non-ATP competitive inhibitor of GSK-3β. At the concentration of 5 mM, LiCl greatly increased the levels of GSK-3β phosphorylated at Ser9 (p-GSK-3β/GAPDH: 0.62 ± 0.057 vs. 0.35 ± 0.012), an inactive form of GSK-3β ( Figure 3A). Treatment with B10 at 10 µM and 20 µM dose-dependently increased the phospho-GSK-3β level in comparison with the control group, and the p-GSK-3β/GAPDH ratio was 0.44 ± 0.037 and 0.53 ± 0.045, respectively.
Elevated levels of hyperphosphorylated tau are highly related to the formation of NFTs in the brains of AD patients. When SH-SY5Y cells were treated with 20 µM Aβ 25-35 for 6 h, the phosphorylation of tau protein at Ser396 increased significantly with a p-tau/GAPDH ratio 0.83 ± 0.061 ( Figure 3C). Treatment with B10 at 5 µM, 10 µM, and 20 µM resulted in a decrease in tau phosphorylation level in a concentration-dependent way (p-tau/GAPDH: 0.65 ± 0.029, 0.40 ± 0.061, and 0.33 ± 0.065). At the concentration of 20 µM, more than half of the hyperphosphorylated tau was reduced by pretreatment with B10.

B10 Selectively Chelates with Fe 2+ , Zn 2+ , Cu 2+ , and Al 3+
Compound B10 was evaluated for its chelating ability with Na + , K + , Mg 2+ , Fe 2+ , Zn 2+ , Ca 2+ , Cu 2+ and Al 3+ by the UV−vis spectroscopy assay [51]. When Na + , K + , Mg 2+ , or Ca 2+ was mixed with B10, the electronic spectra showed no obvious changes ( Figure 4A,B), indicating B10 has little chelating ability with these metal ions. When FeSO 4 was mixed with B10, a red shift from 262 nm to 280 nm were observed, and a new peak appeared at 310 nm ( Figure 4B). In the presence of ZnCl 2 , CuCl 2 , or AlCl 3 , new absorptions at 311, 304, and 302 nm were observed, respectively. These results indicated that B10 could chelate effectively with these ions. There is a controversy over the role of Al 3+ , increasing evidence supports the implication of it in the development of AD [52]. The selective chelating ability of B10 toward Fe 2+ , Zn 2+ , Cu 2+ and Al 3+ makes it as a potential agent targeting metal dyshomeostasis in AD.
Cells 2020, 9, x FOR PEER REVIEW 8 of 15 phosphorylation of tau at Ser396. Protein expression was detected by immunoblot analysis with a specific antibody. Values represent the mean ± SD (n = 3); * p < 0.05, ** p < 0.01 vs. control.
Elevated levels of hyperphosphorylated tau are highly related to the formation of NFTs in the brains of AD patients. When SH-SY5Y cells were treated with 20 μM Aβ25-35 for 6 h, the phosphorylation of tau protein at Ser396 increased significantly with a p-tau/GAPDH ratio 0.83 ± 0.061 ( Figure 3C). Treatment with B10 at 5 μM, 10 μM, and 20 μM resulted in a decrease in tau phosphorylation level in a concentration-dependent way (p-tau/GAPDH: 0.65 ± 0.029, 0.40 ± 0.061, and 0.33 ± 0.065). At the concentration of 20 µ M, more than half of the hyperphosphorylated tau was reduced by pretreatment with B10.

B10 Selectively Chelates with Fe 2+ , Zn 2+ , Cu 2+ , and Al 3+
Compound B10 was evaluated for its chelating ability with Na + , K + , Mg 2+ , Fe 2+ , Zn 2+ , Ca 2+ , Cu 2+ and Al 3+ by the UV−vis spectroscopy assay [51]. When Na + , K + , Mg 2+ , or Ca 2+ was mixed with B10, the electronic spectra showed no obvious changes ( Figure 4A,B), indicating B10 has little chelating ability with these metal ions. When FeSO4 was mixed with B10, a red shift from 262 nm to 280 nm were observed, and a new peak appeared at 310 nm ( Figure 4B). In the presence of ZnCl2, CuCl2, or AlCl3, new absorptions at 311, 304, and 302 nm were observed, respectively. These results indicated that B10 could chelate effectively with these ions. There is a controversy over the role of Al 3+ , increasing evidence supports the implication of it in the development of AD [52]. The selective chelating ability of B10 toward Fe 2+ , Zn 2+ , Cu 2+ and Al 3+ makes it as a potential agent targeting metal dyshomeostasis in AD.
To elucidate why B10 affects Aβ1−42 aggregation and disaggregation, docking simulations were carried out based on the resolved structure of the peptide in its α helix form (PDB code: 1IYT). In

B10 Has a Direct Effect on Aβ 1−42 Aggregation and Disaggregation of Aβ Aggregates and Affects Cu 2+ -Induced Aβ 1−42 Aggregation and Cu 2+ -Aβ 1−42 Aggregates Disaggregation
To determine whether B10 has a direct effect on the inhibition the aggregation of Aβ monomers, we incubated monomeric Aβ 1−42 (20 µL, 40 µM) with B10 (20 µL, 40 µM) and quantified the amount of Aβ aggregates by thioflavin T (ThT) fluorescence assay with curcumin (cur) as a positive control [53]. As shown in Figure 5A, B10 effectively blocked the Aβ aggregate formation with an inhibitory rate of 55.1 ± 3.0%, more potent than curcumin (38.1 ± 6.3%). This was confirmed by TEM images ( Figure 5B). In addition, when B10 was incubated with solutions containing preformed Aβ aggregates, B10 effectively disaggregated the Aβ fibrils, reducing the amount of Aβ aggregates by 61.8 ± 4.1%, much more potent than curcumin (35.3 ± 2.6%) ( Figure 5C). Cells 2020, 9, x FOR PEER REVIEW 9 of 15 interaction [54]. As shown in Figure 5D, B10 interacted with the " 16 KLVFFA 21 " amyloid region. The pyridyl nitrogen atom formed a hydrogen bond with Asp23, and the pyridine ring interacted with Phe20 through π−π stacking. The 4-pyridylamino NH served as a hydrogen bond donor to interact with Glu22. The hydroxyl group in the 8-hydroxyquinoline moiety formed a hydrogen bond with Gln15. All of these interactions may contribute to the direct effect of B10 on Aβ1−42 aggregation and disaggregation. Cu 2+ is a well-known metal ion that modulates the Aβ aggregation and toxicity [42]. To validate whether B10 could target metal dyshomeostasis, the mixtures of Aβ1−42 (20 μL, 40 μM), CuCl2 (20 μL, 40 µ M), and B10 or CQ (40 μL, 40 µ M) were incubated at 37 °C for 24 h and detected by ThT assay. In the presence of Cu 2+ , the fluorescence intensity of the Aβ1−42 and Cu 2+ mixture significantly increased over 1.2-fold in comparison with that of Aβ1−42 alone. As shown in Figure 5E, the positive control CQ, a known nonspecific copper/zinc chelator that can reduce Aβ deposits and improve learning and memory capacities of APP transgenic mice [41], inhibited the Cu 2+ -induced Aβ1−42 aggregation by 66.5 ± 0.9%. B10 was more potent than CQ with an inhibitory rate of 74.6 ± 3.5%.
As an effective agent targeting AD copper homeostasis, the chelator is required not only to inhibit Cu 2+ -induced Aβ aggregation but also to have the ability to extract the copper ion from the aggregates and promote their disaggregation. To evaluate the activity of B10 to disaggregate Cu 2+induced Aβ1−42 aggregation, Aβ1−42 and Cu 2+ were first incubated at 37 °C for 24 h to form Aβ1−42 fibrils, B10 or CQ was then added and incubated for another 24 h. As shown in Figure 5F, B10 showed a similar effect as that of CQ, reducing the Cu 2+ -Aβ1−42 oligomers by 60.3 ± 3.5%. To elucidate why B10 affects Aβ 1−42 aggregation and disaggregation, docking simulations were carried out based on the resolved structure of the peptide in its α helix form (PDB code: 1IYT). In Aβ 1−42 monomer, the self-recognition hydrophobic core " 16 KLVFFA 21 " is located in the central region of the Aβ peptide and is known as the key amyloidogenic sequence that initiates the Aβ−Aβ interaction [54]. As shown in Figure 5D, B10 interacted with the " 16 KLVFFA 21 " amyloid region. The pyridyl nitrogen atom formed a hydrogen bond with Asp23, and the pyridine ring interacted with Phe20 through π−π stacking. The 4-pyridylamino NH served as a hydrogen bond donor to interact with Glu22. The hydroxyl group in the 8-hydroxyquinoline moiety formed a hydrogen bond with Gln15. All of these interactions may contribute to the direct effect of B10 on Aβ 1−42 aggregation and disaggregation.
As an effective agent targeting AD copper homeostasis, the chelator is required not only to inhibit Cu 2+ -induced Aβ aggregation but also to have the ability to extract the copper ion from the aggregates and promote their disaggregation. To evaluate the activity of B10 to disaggregate Cu 2+ -induced Aβ 1−42 aggregation, Aβ 1−42 and Cu 2+ were first incubated at 37 • C for 24 h to form Aβ 1−42 fibrils, B10 or CQ was then added and incubated for another 24 h. As shown in Figure 5F, B10 showed a similar effect as that of CQ, reducing the Cu 2+ -Aβ 1−42 oligomers by 60.3 ± 3.5%.
3.6. B10 Promotes Neuronal Neurite Outgrowth and Growth-Associated Protein 43 (GAP43), N-myc, and Microtubule-Associated Protein 2 (MAP-2) Expressions in SH-SY5Y Cells Since differentiated SH-SY5Y cells possess similar morphology and biochemical processes to mature neurons, they were widely used to study neuronal activity [55]. In cell viability assay, B10 showed no cytotoxicity towards SH-SY5Y cells up to 25 µM ( Figure 6A). B10 was further evaluated for its ability to induce neurogenesis in SH-SY5Y cells.
Cells 2020, 9, x FOR PEER REVIEW 10 of 15 Since differentiated SH-SY5Y cells possess similar morphology and biochemical processes to mature neurons, they were widely used to study neuronal activity [55]. In cell viability assay, B10 showed no cytotoxicity towards SH-SY5Y cells up to 25 μM ( Figure 6A). B10 was further evaluated for its ability to induce neurogenesis in SH-SY5Y cells. There are several recognized biomarkers for neurogenesis, and among them, the growthassociated protein 43 (GAP-43) plays a role at synaptic level [56] and the N-myc gene is essential for normal neurogenesis and regulates cell proliferation, differentiation, and nuclear size [57], while the microtubule-associated protein 2 (MAP-2) is abundant in the mammalian nervous system and associated with the neurites and dendrite scaffold formation [58]. RA is a vitamin A metabolite that is widely used in the differentiation of SH-SY5Y cells into highly homogeneous populations of neuron-like cells [59]. After the treatment of SH-SY5Y cells with B10 or RA (10 μM) for 24 h, the mRNA expression of GAP43, N-myc, and MAP-2 was assessed by RT-PCR analysis. As shown in Figure 6B, RA induced significant expression of all the three markers. In comparison with RA, B10 exhibited more profound effects and was more than 2-fold more active than RA in promotion of the expression of GAP43, N-myc, and MAP-2 in SH-SY5Y cells. To confirm the obtained results, the morphology of the differentiated neuronal neurite outgrowth was assessed after cell treatment for 72 h with B10 or RA (10 μM). In agreement with the RT-PCR results, B10 induced more significant neurite outgrowth than RA ( Figure 6C).

Discussion
Due to the complexity of AD pathogenesis, there is no effective treatment to prevent or reverse the progression of AD, and the discovery of effective anti-AD drugs is rather challenging. In recent years, the design of a multifunctional inhibitor that target multifacet of AD was a well-accepted strategy. In this study, we designed a novel GSK-3β inhibitor B10 as a multitarget-directed ligand through incorporating the 8-hydroxyquinoline motif in the molecule to target AD metal There are several recognized biomarkers for neurogenesis, and among them, the growth-associated protein 43 (GAP-43) plays a role at synaptic level [56] and the N-myc gene is essential for normal neurogenesis and regulates cell proliferation, differentiation, and nuclear size [57], while the microtubule-associated protein 2 (MAP-2) is abundant in the mammalian nervous system and associated with the neurites and dendrite scaffold formation [58]. RA is a vitamin A metabolite that is widely used in the differentiation of SH-SY5Y cells into highly homogeneous populations of neuron-like cells [59]. After the treatment of SH-SY5Y cells with B10 or RA (10 µM) for 24 h, the mRNA expression of GAP43, N-myc, and MAP-2 was assessed by RT-PCR analysis. As shown in Figure 6B, RA induced significant expression of all the three markers. In comparison with RA, B10 exhibited more profound effects and was more than 2-fold more active than RA in promotion of the expression of GAP43, N-myc, and MAP-2 in SH-SY5Y cells. To confirm the obtained results, the morphology of the differentiated neuronal neurite outgrowth was assessed after cell treatment for 72 h with B10 or RA (10 µM). In agreement with the RT-PCR results, B10 induced more significant neurite outgrowth than RA ( Figure 6C).

Discussion
Due to the complexity of AD pathogenesis, there is no effective treatment to prevent or reverse the progression of AD, and the discovery of effective anti-AD drugs is rather challenging. In recent years, the design of a multifunctional inhibitor that target multifacet of AD was a well-accepted strategy. In this study, we designed a novel GSK-3β inhibitor B10 as a multitarget-directed ligand through incorporating the 8-hydroxyquinoline motif in the molecule to target AD metal dyshomeostasis.
GSK-3β is an important target for the treatment of AD because it acts as a linker between the two important histopathological hallmarks: NFTs formed by the accumulation of hyper-phosphorylated tau protein, and the extracellular senile plaques caused by abnormal Aβ aggregation. The overactivation of GSK-3β is highly related to the abnormal tau hyperphosphorylation that plays a crucial role in axonal assembly associated with synaptic transmission, neuronal degeneration and dysfunction, and NFTs. B10 was proven to be a potent GSK-3β inhibitor. It effectively increased the inactivated form of GSK-3β through the phosphorylation at Ser9 and reduced the phosphorylation of tau protein at Ser396 in SH-SY5Y cells.
Aβ protein, the main component of senile plaques, is a hydrophobic, intrinsically disordered peptide with 36−43 amino acids [60]. Aβ is generated by the sequential metabolism of amyloid precursor protein by βand γ-secretases, and Aβ 1−42 is the most toxic and predominant species found in plaques [61]. Soluble monomeric Aβ 1−42 is prone to self-assembly to form Aβ oligomers that aggregate to form protofibrils and then mature amyloid fibrils. There is increasing evidence that the soluble prefibrillar Aβ oligomers are more toxic than mature fibrils [62]. Therefore, an ideal anti-AD drug should have the ability to prevent the formation as well as to disaggregate the Aβ oligomers and fibrils. Though GSK-3β has some effect on Aβ formation through its regulation of γ-secretase [32], the inhibition of GSK-3β may have limited effects on reducing the Aβ level, considering the complicated processes in Aβ formation and its clearance. In this regard, B10 is unique as a potent GSK-3β inhibitor that can inhibit Aβ oligomers formation and promote Aβ oligomers and fibrils disaggregation. As far as we know, B10 is the first potent GSK-3β inhibitor reported to play a direct role in Aβ. Molecular docking simulations revealed the binding mode of B10 with Aβ 1−42 in the self-recognition hydrophobic " 16 KLVFFA 21 " motif.
The homeostasis dysregulation of transition metal ions (Cu 2+ , Zn 2+, and Fe 2+ ), especially Cu 2+ , not only promotes the aggregation and deposition of Aβ, but also generates ROS, leading to oxidative stress [41,42]. Metal chelating agents, such as CQ, can prevent or reverse copper-Aβ interactions and reduce plaque burden in the brain of transgenic AD mouse models with marked improvements in cognitive performance [63,64]. Clinical investigations proved that CQ or its derivative could improve cognitive scores and reduce Aβ levels in AD patients [65,66]. These studies suggest that targeting copper dyshomeostasis provides a choice for AD therapy. By introducing an 8-hydroxyquinoline moiety as a bidentate chelator, B10 selectively chelates with AD-related metal ions, inhibits Cu 2+ -induced Aβ 1−42 aggregation, and promotes the disaggregation of Cu 2+ -Aβ 1−42 oligomers, serving as a potent ligand to target AD metal dyshomeostasis.
A recent study confirms that new neurons are born throughout aging in healthy humans, but this drops sharply in AD patients, suggesting that therapeutic strategies aimed at increasing neurogenesis may slow the disease [67]. β-Catenin is a substrate of GSK-3β. Phosphorylation of β-catenin in the absence of Wnt signaling by GSK3β results in its ubiquitination and subsequent degradation by proteasomes [68]. As a necessary transcriptional coactivator, β-catenin not only influences cellular events but also plays important roles in cell adhesion complexes, including those necessary for neuronal differentiation. The depletion of β-catenin by small interfering RNA blocked the neurite outgrowth in SH-SY5Y cells [69]. As a potent GSK-3β inhibitor, B10 effectively increases the β-catenin abundance, promotes neuronal neurite outgrowth, and induces significant expression of GAP43, N-myc, and MAP-2 in SH-SY5Y cells. These results are probably a combined effect of B10 on the GSK-3β and β-catenin signal pathways.

Conclusions
Based on the GSK-3β crystal structure and through rational drug design and molecule docking simulations, we designed and screened novel pyrrolo [2,3-b]pyridine-based GSK-3β inhibitors with the incorporation of 8-hydroxyquinoline motif as a bidentate ligand to target AD metal dyshomeostasis as well. Compound B10 was identified as a potent GSK-3β inhibitor. Unlike other GSK-3β inhibitors reported, B10 is unique and has a direct effect on Aβ, inhibiting the Aβ oligomers formation and promoting Aβ oligomers and fibrils disaggregation. B10 also targets tau and metal dyshomeostasis. Moreover, B10 is more potent than RA in promoting neuronal neurite outgrowth and in inducing GAP43, N-myc, and MAP-2 expressions in SH-SY5Y cells, possibly through the GSK-3β and β-catenin signal pathways. Therefore, B10 could serve as a good lead for future structural optimization and further in vivo investigations as a multifunctional agent for the treatment of AD.