Exploration of multi‐target effects of 3‐benzoyl‐5‐hydroxychromen‐2‐one in Alzheimer’s disease cell and mouse models

Abstract Microtubule‐associated protein Tau, abundant in the central nervous system (CNS), plays crucial roles in microtubule assembly and stabilization. Abnormal Tau phosphorylation and aggregation are a common pathogenic hallmark in Alzheimer's disease (AD). Hyperphosphorylation of Tau could change its conformation and result in self‐aggregation, increased oxidative stress, and neuronal death. In this study, we examined the potential of licochalcone A (a natural chalcone) and five synthetic derivatives (LM compounds) for inhibiting Tau misfolding, scavenging reactive oxygen species (ROS) and providing neuroprotection in human cells expressing proaggregant ΔK280 TauRD‐DsRed. All test compounds were soluble up to 100 μM in cell culture media and predicted to be orally bioavailable and CNS‐active. Among them, licochalcone A and LM‐031 markedly reduced Tau misfolding and associated ROS, promoted neurite outgrowth, and inhibited caspase 3 activity in ΔK280 TauRD‐DsRed 293 and SH‐SY5Y cells. Mechanistic studies showed that LM‐031 upregulates HSPB1 chaperone, NRF2/NQO1/GCLC pathway, and CREB‐dependent BDNF/AKT/ERK/BCL2 pathway in ΔK280 TauRD‐DsRed SH‐SY5Y cells. Decreased neurite outgrowth upon induction of ΔK280 TauRD‐DsRed was rescued by LM‐031, which was counteracted by knockdown of NRF2 or CREB. LM‐031 further rescued the downregulated NRF2 and pCREB, reduced Aβ and Tau levels in hippocampus and cortex, and ameliorated cognitive deficits in streptozocin‐induced hyperglycemic 3 × Tg‐AD mice. Our findings strongly indicate the potential of LM‐031 for modifying AD progression by targeting HSPB1 to reduce Tau misfolding and activating NRF2 and CREB pathways to suppress apoptosis and promote neuron survival, thereby offering a new drug development avenue for AD treatment.

examined the potential of licochalcone A (a natural chalcone) and five synthetic derivatives (LM compounds) for inhibiting Tau misfolding, scavenging reactive oxygen species (ROS) and providing neuroprotection in human cells expressing proaggregant ΔK280 Tau RD -DsRed. All test compounds were soluble up to 100 μM in cell culture media and predicted to be orally bioavailable and CNS-active. Among them, licochalcone A and LM-031 markedly reduced Tau misfolding and associated ROS, promoted neurite outgrowth, and inhibited caspase 3 activity in ΔK280 Tau RD -DsRed 293 and SH-SY5Y cells. Mechanistic studies showed that LM-031 upregulates HSPB1 chaperone, NRF2/NQO1/GCLC pathway, and CREB-dependent BDNF/AKT/ERK/BCL2 pathway in ΔK280 Tau RD -DsRed SH-SY5Y cells. Decreased neurite outgrowth upon induction of ΔK280 Tau RD -DsRed was rescued by LM-031, which was counteracted by knockdown of NRF2 or CREB. LM-031 further rescued the downregulated NRF2 and pCREB, reduced Aβ and Tau levels in hippocampus and cortex, and ameliorated cognitive deficits in streptozocin-induced hyperglycemic 3 × Tg-AD mice. Our findings strongly indicate the potential of LM-031 for modifying AD progression by targeting HSPB1 to reduce Tau misfolding and activating NRF2 and CREB pathways to suppress apoptosis and promote neuron survival, thereby offering a new drug development avenue for AD treatment.

| INTRODUC TI ON
Several neurodegenerative disorders known as tauopathies, including Alzheimer's disease (AD), are characterized by abnormal aggregation of Tau protein (Iqbal et al., 2005). In these diseases, Tau becomes abnormally hyperphosphorylated and misfolded, forming insoluble aggregates. Phosphorylation of Tau has been proposed as the link between oxidative stress, mitochondrial dysfunction, and connectivity failure (Mondragón-Rodríguez et al., 2013). Chaperones and the ubiquitin-proteasome system, playing a defense role to remove misfolded proteins and rescue the neurons, are involved particularly in the early stage of tauopathies (Sherman & Goldberg, 2001). Thus, identification of potential Tau misfolding inhibitors could be a strategy for disease progression modification in these neurodegenerative diseases.
Encoded on chromosome 17, Tau is mainly expressed in neuronal axons, where they play an important role in microtubule dynamics and assembly, as well as in axonal transport and apoptosis (Lee, Goedert, & Trojanowski, 2001). Tau is a prototypical natively unfolded protein (Schweers, Schonbrunnhanebeck, Marx, & Mandelkow, 1994).
Tau binds microtubules through C-terminal repeat domain (Tau RD ), which contains four highly conserved 18-amino acid repeat domains (Goedert, Spillantini, Potier, Ulrich, & Crowther, 1989). The Tau RD has been shown to reside at the core of paired helical filaments of neurofibrillary tangles (Goedert et al., 1989), one of the neuropathological hallmarks of AD, and a deletion mutant ∆K280 within Tau RD (∆K280 Tau RD ) has been found in patients with AD and other tauopathies (D'Souza et al., 1999;Momeni et al., 2009;Rizzu et al., 1999). Overexpressed ∆K280 Tau RD in mouse neuroblastoma N2a cells is highly prone to being misfolded and forming aggregation (Khlistunova et al., 2006). Chemicals that prevent misfolding of Tau in the ∆K280 Tau RD -expressing N2a cells are thought to be candidates for the treatment of AD and tauopathies (Pickhardt et al., 2005(Pickhardt et al., , 2007. Heat shock protein system that helps refolding of abnormal protein is also emerging as a key pathway for modulation to prevent AD-related neurodegenerative damage (Trovato, Siracusa, Di Paola, Scuto, Fronte, et al., 2016;Trovato, Siracusa, Di Paola, Scuto, Ontario, et al., 2016). We have also established human cells expressing pro-aggregated ∆K280 Tau RD to screen herbal extracts or indole compounds exerting neuroprotection for AD and tauopathies treatment (Chang et al., 2016(Chang et al., , 2017Chen et al., 2018).
Here, we examined the ability of licochalcone A and five derivative compounds to prevent ∆K280 Tau RD aggregation and oxidation as well as to promote neuroprotection by using human cells expressing DsRed-tagged ∆K280 Tau RD (Chang et al., 2016). We also explored the potential of LM-031 for AD treatment in streptozocin (STZ)induced hyperglycemic 3 × Tg-AD mice. Our findings indicate that the synthetic LM-031 is a potential candidate for the development of AD and tauopathy therapeutics.

| Test compounds and IC 50 cytotoxicity
Licochalcone A and five derivative LM compounds were tested ( Figure 1a). All six compounds were soluble up to 100 μM in a cell culture medium ( Figure S1a). Based on molecular weight (MW), hydrogen bond donors (HBD), hydrogen bond acceptors (HBA), and calculated octanol/water partition coefficient (cLogP), all six compounds meet Lipinski's criteria in predicting oral bioavailability

| Reduction of Tau misfolding, ROS, and caspase 3 activity of licochalcone A and LM-031 in ∆K280 Tau RD -DsRed-expressing 293 cells
In ∆K280 Tau RD -DsRed 293 cells, a proaggregant (ΔK280) Tau repeat domain (Tau RD ) containing residues 244-372 of the full-length 4R2N Tau441 was fused in-frame with DsRed fluorophore. Compared to the wild-type Tau RD , the poorly folded ΔK280 Tau RD formed aggregates, which adversely affected the folding of fused DsRed and thus decreased DsRed fluorescence, as also evidenced by increase in thioflavin S fluorescence emission (Chang et al., 2017). Thioflavin S is a reporter dye for studies of Tau toxicity and pharmacologic prevention in cell model of tauopathy (Pickhardt et al., 2017). It binds to amyloid fibrils but not monomers and gives a distinct increase in fluorescence emission. Inhibition of Tau aggregation may improve DsRed misfolding, leading to increased fluorescence in ΔK280 Tau RD -DsRed-expressing cells. Utilizing the established Tet-on ΔK280 Tau RD -DsRed 293 cells (Chang et al., 2016), licochalcone A and LM compounds were tested for effects of reducing Tau misfolding and anti-oxidation ( Figure S2a). Congo red was included for comparison.

| Promotion of neurite outgrowth and reduction of caspase 3 activity of licochalcone A and LM-031 in ∆K280 Tau RD -DsRed-expressing SH-SY5Y cells
Since licochalcone A or LM-031 at 1 µM has good effects in antioxidation and anti-aggregation, the neuroprotective effects of licochalcone A or LM-031 were evaluated using ∆K280 Tau RD -DsRed-expressing SH-SY5Y cells ( Figure S4a). As shown in Figure 1c
In addition to HSPB1 chaperone, we examined the effects of licochalcone A and LM-031 on expression levels of NRF2, CREB, and downstream targets (Andrisani, 1999;Zhang et al., 2013)

| PAMPA to assess BBB permeability of LM-031
To date, parallel artificial membrane permeability assay (PAMPA) models that exhibit a high degree of correlation with permeation across a variety of barriers, including BBB (Di, Kerns, Fan, McConnell, & Carter, 2003;Ottaviani, Martel, Escarala, Nicolle, & Carrupt, 2008), have been developed. We then used PAMPA-BBB method to estimate the passive BBB permeability of LM-031. Quality control compounds, testosterone (high-permeability marker), theophylline (low-permeability marker), and Lucifer yellow (integrity marker), were included for comparison. The effective permeability (P e ) values of testosterone and theophylline were 21.75 and 0.11 (10 -6 cm/s), respectively, representing high (>4 × 10 -6 cm/s) and low (<2 × 10 -6 cm/s) BBB permeable controls ( Figure S6). The BBB permeability values of testosterone and theophylline were similar to the previously published results (Di et al., 2003). The transport of Lucifer yellow was undetectable, indicating good membrane integrity (below the 0.1% cut off). The P e value of LM-031 was 4.80 ± 0.12 (10 -6 cm/s), suggesting that LM-031 could be categorized as a high BBB permeable compound (P e > 4 × 10 -6 cm/s) in PAMPA-BBB measurement. F I G U R E 2 Enhanced expression of HSPB1, NRF2, and CREB pathways following LM-031 administration in ΔK280 Tau RD -DsRed SH-SY5Y cells. Relative (a) HSPB1 and soluble Tau RD -DsRed, (b) NRF2, GCLC, and NQO1, as well as (c) CREB, pCREB, BCL2, BAX, GADD45B, and BDNF protein levels were analyzed through immunoblotting using specific antibodies. Protein levels were normalized to β-actin internal control. Relative protein levels are shown on the right side of the representative Western blot images. The relative protein level in uninduced cells was normalized (100%). For Tau RD -DsRed, the soluble level in induced cells was set at 100%. In addition, ΔK280 Tau RD -DsRed mRNA levels (a) were quantitated by real-time PCR (n = 3). p values: comparisons between induced and uninduced cells ( # : p < .05 and ## : p < .01), or between treated and untreated cells (*: p < .05, **: p < .01, and ***: p < .001) (n = 3). (one-way ANOVA with a post hoc Tukey test)

| LM-031 ameliorated spatial learning and memory deficit in STZ-treated 3 × Tg-AD mice
3 × Tg-AD mice were used as an in vivo animal model to explore the potential of LM-031 for AD treatment. At 6 months of age, the homozygous 3 × Tg-AD mice displayed extracellular β-amyloid plaques and intraneuronal Aβ buildup in the brain (Oddo et al., 2003), accompanied by more training to learn the Morris water maze task (Billings, Oddo, Green, McGaugh, & LaFerla, 2005).
Intraneuronal aggregates of conformationally altered hyperphosphorylated Tau were not evident until the 3 × Tg-AD mice reach about 12 months of age (Oddo et al., 2003). As impairment of brain glucose metabolism correlates well with the clinical symptoms in AD (Mosconi, 2005) and STZ exacerbated cognitive deficits in 3 × Tg-AD mice (Chen, Liang, et al., 2014), STZ-induced hyperglycemia was applied by i.p. injections of 4 STZ doses to 6-month-old 3 × Tg-AD mice to accelerate the development of AD phenotypes 107-112 mg/dl) (p < .001). These results showed that STZ induced a hyperglycemic condition in 6-month-old 3 × Tg-AD mice, which F I G U R E 3 NRF2 and CREB as therapeutic targets in LM-031-treated ΔK280 Tau RD -DsRed SH-SY5Y cells. (a) Western blot analysis of CREB, pCREB, cleaved CASP3, and Tau RD -DsRed levels in compound-treated cells infected with CREB-specific or scrambled shRNAexpressing lentivirus. GAPDH was used as a loading control. To normalize, the relative CREB, pCREB, or CASP3 of uninduced cells, or relative Tau RD -DsRed of untreated cells was set at 100%. (b) Western blot analysis of NRF2 and Tau RD -DsRed levels (loading control: GAPDH) and caspase 3 activity assay in compoundtreated cells infected with NRF2-specific or scrambled shRNA-expressing lentivirus. To normalize, the relative NRF2 level or caspase 3 activity of uninduced cells, or relative Tau  An open-field test was conducted on day 24 to explore the spontaneous motor activities and anxious mood in mice. No significant change was observed in distance traveled and inactive time of 3 × Tg-AD mice with or without STZ or LM-031 ( Figure S7b). On day 26, Y-maze alternation task based on the natural tendency of the mice to explore novel environment was performed to assess spatial working memory. As shown in Figure 4b, Y-maze alternation rate was reduced in hyperglycemic (STZ) 3 × Tg-AD mice as compared to normoglycemic group (-STZ) (53% versus 63%; p = .029), and LM-031 treatment (STZ/LM-031 group) significantly increased alternation rate (from 53% to 72%; p < .001). The results suggest enhanced working memory in LM-031-treated hyperglycemic 3 × Tg-AD mice.
To assess the effect of LM-031 on hippocampus-dependent learning and memory retrieval, we conducted Morris water maze task in mice with training (days 30-33), testing (day 34), and probe In the probe trials on days 34 and 36, the duration in target quadrant of the hyperglycemic mice was markedly less than that of the normoglycemic mice (17 s versus 26 s; p = .011-0.006), and LM-031 treatment extended the time spent in the target quadrant (from 17 to 24 s; p = .060-0.047) for both 2 and 48 hr trials. Thus, LM-031 treatment increased the retrieval of spatial memory in hyperglycemic 3 × Tg-AD mice.

| LM-031 increased NeuN level and decreased Aβ and Tau levels in STZ-treated 3 × Tg-AD mice
In addition to cognitive function, we examined NeuN (RNA binding protein, fox-1 homolog (C. elegans) 3), Aβ, and Tau levels in 3 × Tg-AD mice with or without STZ or LM-031 treatment.
However, the relative total CREB level was not significantly different among the three groups.

| D ISCUSS I ON
In this study, we demonstrated Tau misfolding reduction, anti-oxidation, and neuroprotection effects of in-house synthetic compound LM-031. Treatment with LM-031 increases soluble ∆K280 Tau RD -DsRed protein and fluorescence intensity without changing the Tau RD -DsRed RNA level (Figures 1b, 2a), which reflects that the increased fluorescence intensity is mainly due to the improvement of Tau misfolding state but not caused by increased Tau RD -DsRed transcription. In other words, neurite outgrowth-promoting effect of LM-031 in our tauopathy model is probably due to the amelioration of tau misfolding, which leads to reduction of oxidative stress and caspase 3 activity (Figure 1). Further target identification from mechanistic studies indicates that LM-031 upregulates HSPB1, NRF2, and CREB expression in differentiated ΔK280 Tau RD -DsRed SH-SY5Y cells to suppress apoptosis and promote neuron survival (Figures 2, 3).
The different indolylquinoline (Chang et al., 2017) and chalcone (Lee et al., 2018 and this study) compounds with similar effect on enhancing HSPB1 expression are not uncommon, as compounds may bind to different promotor sites to enhance HSPB1 expression. We also proposed another hypothesis that one of them may act on upstream gene that controls the HSPB1 expression and the other directly on HSPB1.
Oxidative stress has been recognized as a contributing factor to aging and progression of multiple neurodegenerative diseases including AD. Increased production of ROS and reduced antioxidant defense could affect synaptic activity leading to cognitive dysfunction, and accumulation of hyperphosphorylated Tau protein could exacerbate ROS production (Tönnies & Trushina, 2017). The levels of HSPs correlate inversely with the levels of granular Tau oligomers in human brain, implicating that heat shock proteins may regulate soluble Tau protein levels to counteract the formation of granular Tau oligomers (Sahara et al., 2007). By activating HSPB1 expression to reduce aggregation, ROS and cytotoxicity in ΔK280 Tau RD -DsRed-expressing cells (Figure 1), LM-031 could be a promising disease-modifying compound for the treatment of AD and other oxidative stress-related neurodegenerative diseases.
Among the mechanisms of protection against oxidative stress, activation of NRF2 increases levels of phase 2 antioxidant enzymes such as NQO1 and GCLC (Zhang et al., 2013). A significant decrease in nuclear NRF2 levels observed in postmortem brains of human AD cases suggests that NRF2-mediated transcription is not activated in neurons in AD despite the presence of oxidative stress (Ramsey et al., 2007). Oxidative stress involving changes in NRF2 constitutes one of the early events in 3 × Tg-AD mice (Mota et al., 2015). By combining the NRF2 activator 18α-glycyrrhetinic acid and the NADH precursor nicotinamide, neuronal survival in 3 × Tg-AD against β-amyloid stress was additively increased (Ghosh, LeVault, & Brewer, 2014). In addition, carnosic acid, a compound that is activated by oxidative stress to stimulate NRF2 transcription pathway, exhibits neuroprotective effects both in vitro and in transgenic AD mice (Lipton et al., 2016). In our study, treatment with LM-031 upregulated antioxidant NRF2 pathway and reduced proapoptotic BAX and CASP3 expression in pro-aggregated Tau RD -DsRed-expressing SH-SY5Y cells, supporting the role of LM-031 in treating AD by preventing oxidative stress and blocking apoptosis pathway.
Synaptic dysfunction in AD has been a recent focus as memory formation is accompanied by altered synaptic strength and synaptic changes are highly correlated with the severity of clinical dementia (Teich et al., 2015). It is well known that CREB plays a crucial role in memory consolidation (Silva, Kogan, Frankland, & Kida, 1998), as phosphorylated CREB binds to CBP (CREB-binding protein, a histone acetyltransferase that catalyzes histone acetylation), causing an increase in the transcription of memory-associated genes (Kandel, 2004;Korzus, Rosenfeld, & Mayford, 2004). CREB downregulation has been implicated in the pathology of AD (Bartolotti, Bennett, & Lazarov, 2016;Pugazhenthi, Wang, Pham, Sze, & Eckman, 2011;Yamamoto-Sasaki, Ozawa, Saito, Rosler, & Riederer, 1999). Overexpression of human wild-type full-length Tau in hippocampal neurons impaired synapse impairment and memory deficits by dephosphorylation of CREB mediated through upregulating calcium/calmodulin-dependent protein phosphatase calcineurin and suppressing calcium/calmodulin-dependent protein kinase IV (Yin et al., 2016). Thus, increasing the expression of CREB could be considered as a possible therapeutic target for In this study, the therapeutic potential of LM-031 in AD was further examined in 3 × Tg-AD mice which displayed the pathology and synaptic deficits of AD in an age-dependent manner (Oddo et al., 2003). AD occurs as a result of complex interactions be-  (Wang et al., 2014). Therefore, hyperglycemia was used in this study to accelerate the phenotypes of 3 × Tg-AD mice to assess therapeutic efficacy of LM-031. In vivo, the observed beneficial effects of LM-031 on learning and memory improvement in hyperglycemic 3 × Tg-AD mice (Figure 4) could be mediated by upregulating NRF2 and pCREB and reducing Aβ and Tau levels in hippocampus and cortex of the hyperglycemic 3 × Tg-AD mice (Figures 5, 6).
Both Aβ and ∆K280 Tau RD attenuate the expression level of CREB in SH-SY5Y cells (Lee et al., 2018 and Figure 2c). However, no apparent reduction of CREB protein level was observed in STZinduced hyperglycemic 3 × Tg-AD mice (Figure 6). Although a previous study of brain gene expression of 3 × Tg-AD mice revealed the reduced expression of CREB, the reduction did not reach statistical significance (Chen et al., 2012). CBP gene transfer to 3 × Tg-AD mice increased the reduced pCREB to increase BDNF levels and ameliorate learning and memory deficits without affecting total CREB level (Caccamo, Maldonado, Bokov, Majumder, & Oddo, 2010). In intracerebroventricular STZ-induced memory-impaired rats, STZ injection significantly downregulated the pCREB but not CREB expression in the cerebral cortex and hippocampus F I G U R E 6 LM-031 augments NFR2 and pCREB expression in STZ-treated 3 × Tg-AD mice. Expression levels of NRF2, pCREB, and CREB in hippocampus were analyzed by Western blot using GAPDH as a loading control. To normalize, the relative NRF2, pCREB, and CREB of, STZ mice was set as 100%. p values, STZ versus -STZ mice or STZ/LM-031 versus STZ mice. *: p < .05, and ## and **: p < .01. (one-way ANOVA with a post hoc Tukey test) (Rajasekar, Nath, Hanif, & Shukla, 2017). In STZ-induced diabetic mice, pCREB/CREB and the expression of BDNF protein in the hippocampus were obviously decreased in STZ group, without affecting CREB level (Tang et al., 2018). In accordance with the previous studies using AD or STZ-induced animal models, our study results suggest that the main pathology is reduced pCREB rather than total CREB and the compound enhancing pCREB is able to rescue the phenotype and neurodegeneration of STZ-accelerated AD mice.
Currently, no effective treatment exists for modifying or preventing AD progression; thus, new strategies are desperately needed. Tau misfolding and aggregation cause oxidative stress and associated neuronal injury and cell death. Taken together, in-house synthetic LM-031 exerted neuroprotective effects in ΔK280 Tau RD -DsRed SH-SY5Y cells by upregulating molecular chaperone HSPB1 and NRF2/NQO1/GCLC to reduce oxidative stress and by activating CREB-dependent BDNF/AKT/ERK and BCL2 for cell survival and anti-apoptosis. With low cytotoxicity in human cells, high solubility in cell culture medium, high BBB permeability, and prediction of orally bioavailable, LM-031 has potential for being developed as an AD therapeutic. Additional preclinical studies are required before applying this multi-target compound to modify the progression of AD in clinical trials.

| MATERIAL S AND ME THODS
In-house LM compounds were synthesized and characterized by NMR spectrum. Tet-On ∆K280 Tau  Reporting In Vivo Experiments) guidelines. Open-field task, Y-maze task, Morris water maze task, and immunohistochemistry and image analysis were performed.
Full methods are available in the Supporting Information.

ACK N OWLED G M ENTS
We thank the Molecular Imaging Core Facility of National Taiwan Normal University for the technical assistance. This work was supported by the grants 107-2320-B-003-006, 107-2320-B-182A-020, and 107-2811-B-003-506 from the Ministry of Science and Technology, Taiwan.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
GJLC and CMC designed the study, analyzed data, and wrote the manuscript. THL and YJC performed most of the experimental work and analyzed data. CHL, CYL, CYC, YCC, and SMY performed some experiments. WL, HMHL, YRW, and KHC reviewed the manuscript.
All authors approved the final version of the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available in Mendeley Data at https://data.mende ley.com/datas ets/f78fj p74c2 /draft ?a=c4977 b15-e804-40de-8425-fa9c2 9594a26.