Combined Treatment with Curcumin and Ferulic Acid Suppressed the Aβ-Induced Neurotoxicity More than Curcumin and Ferulic Acid Alone

Alzheimer’s disease (AD) is a neurodegenerative disease that leads to progressive cognitive decline. Several effective natural components have been identified for the treatment of AD. However, it is difficult to obtain conclusive evidence on the safety and effectiveness of natural components, because a variety of factors are associated with the progression of AD pathology. We hypothesized that a therapeutic effect could be achieved by combining multiple ingredients with different efficacies. The purpose of this study was thus to evaluate a combination treatment of curcumin (Cur) and ferulic acid (FA) for amyloid-β (Aβ)-induced neuronal cytotoxicity. The effect of Cur or FA on Aβ aggregation using thioflavin T assay was confirmed to be inhibited in a concentration-dependent manner by Cur single or Cur + FA combination treatment. The effects of Cur + FA on the cytotoxicity of human neuroblastoma (SH-SY5Y) cells induced by Aβ exposure were an increase in cell viability, a decrease in ROS and mitochondrial ROS, and repair of membrane damage. Combination treatment showed an overall higher protective effect than treatment with Cur or FA alone. These results suggest that the combined action mechanisms of Cur and FA may be effective in preventing and suppressing the progression of AD.


Introduction
Alzheimer's disease (AD), an age-related neurodegenerative disorder that causes progressive cognitive decline, represents the most frequent form of dementia. The "Alzheimer's Association Report 2021" predicts that more than 152 million people will have AD by 2050 [1]. One of the neuropathological features of AD is the deposition of senile plaques and neurofibrillary changes in the brain. The significant components of senile plaques and neurofibrillary changes have been identified as amyloid-β protein (Aβ) and highly phosphorylated tau protein. These protein aggregates induce neuronal damage and cell death, resulting in memory and learning disabilities that lead to dementia. The exact cause of AD is not yet clear, but a number of studies have shown that oxidative stress, mitochondrial dysfunction, and inflammation play a major role in its onset and progression [2]. The amyloid cascade hypothesis has also been a major theoretical component of research on AD for over 20 years [3]. Importantly, Aβ begins to accumulate in the brain 10 to 20 years or more before the cognitive decline is observed [4], and there is therefore an urgent need

Effects of Cur, FA, and Their Combination on the Aggregation of Aβ1-40 and Aβ1-42
We compared the effects of Cur, FA, and the combination of both on the aggre kinetics of Aβ1-40 and Aβ1-42 peptides using a thioflavin T (ThT) fluorescence assay we monitored the amyloid formation of Aβ1-40 (25 μM) (Figure 2A,B). The aggregat the Aβ1-40 peptide alone increased exponentially with no delay and reached ap mately twice the fluorescence intensity observed at onset after 6 h. However, in the ence of Cur, the fluorescence intensity of ThT decreased in a concentration-depe manner as compared to peptide incubated alone, with 10 μM Cur instigating an inhi of 55% after 360 min (Aβ1-40 alone: 554,963. 3 Table S1).
Next, we monitored the aggregation kinetics of Aβ1-42 peptide for 120 min (F 2B). The aggregation of Aβ1-42 peptide causes significant neurotoxicity among all ex isoforms of Aβ [23]. Therefore, we also investigated the inhibitory effects of Cur, FA the combination of both on Aβ1-42 aggregation kinetics, as described above for Aβ1aggregation rate of Aβ1-42 peptide was much faster than that of Aβ1-40 peptide. In a ous study, we had also observed that the aggregation kinetics of Aβ1-42 peptide showed an exponential increase without any lag phase [24]. Here, the first stage w exponential increase in ThT fluorescence intensity, after which the stationary phas reached. The maximum saturation fluorescence intensity value of the stationary ph creased to about 5.5-fold the fluorescence intensity at onset. Similar to the aggre kinetics of the Aβ1-40 peptide, we observed that the incubation of Aβ1-42 peptide wit resulted in a concentration-dependent decrease in fluorescence intensity. After 12 the fluorescence of Aβ1-42 incubated with concentration above 1 μM Cur was signifi lower than that in the absence of Cur (Aβ1-42 alone: 2,518,582.3 ± 27,558.1, 1 μM 2,196,175.5 ± 32,443.2, 5 μM Cur: 1,349,023 ± 15908.3, 10 μM Cur: 788,601.5 ± 66,620 6) ( Figure 2B and Table S2). Around 70% inhibition of Aβ1-42 peptide aggregation w served after incubation with 10 μM Cur for 120 min. However, similar to Aβ1- 40   We compared the effects of Cur, FA, and the combination of both on the aggregation kinetics of Aβ 1-40 and Aβ 1-42 peptides using a thioflavin T (ThT) fluorescence assay. First, we monitored the amyloid formation of Aβ   (25 µM) (Figure 2A Table S1).
Next, we monitored the aggregation kinetics of Aβ 1-42 peptide for 120 min ( Figure 2B). The aggregation of Aβ 1-42 peptide causes significant neurotoxicity among all existing isoforms of Aβ [23]. Therefore, we also investigated the inhibitory effects of Cur, FA, and the combination of both on Aβ 1-42 aggregation kinetics, as described above for Aβ  . The aggregation rate of Aβ 1-42 peptide was much faster than that of Aβ 1-40 peptide. In a previous study, we had also observed that the aggregation kinetics of Aβ 1-42 peptide alone showed an exponential increase without any lag phase [24]. Here, the first stage was an exponential increase in ThT fluorescence intensity, after which the stationary phase was reached. The maximum saturation fluorescence intensity value of the stationary phase increased to about 5.5-fold the fluorescence intensity at onset. Similar to the aggregation kinetics of the Aβ 1-40 peptide, we observed that the incubation of Aβ 1-42 peptide with Cur resulted in a concentration-dependent decrease in fluorescence intensity. After 120 min, the fluorescence of Aβ 1-42 incubated with concentration above 1 µM Cur was significantly lower than that in the absence of Cur (Aβ 1-42 alone: 2,518,582.3 ± 27,558.1, 1 µM Cur: 2,196,175.5 ± 32,443.2, 5 µM Cur: 1,349,023 ± 15908.3, 10 µM Cur: 788,601.5 ± 66,620.7, n = 6) ( Figure 2B and Table S2). Around 70% inhibition of Aβ 1-42 peptide aggregation was observed after incubation with 10 µM Cur for 120 min. However, similar to Aβ 1-40 , no significant effect on Aβ   at 120 min (1 μM Cur + 10 μM FA: 2,150,414.8 ± 7235.5, 5 μM Cur + 10 μM FA: 1,164,776.6 ± 58,882, n = 6) ( Figure 2B and Table S2).  As shown in Figure 3 and Figure S1, to determine the neurotoxicity of Aβ1-42 and to compare the effects of Cur, FA, and their combination on cells, the viability of SH-SY5Y cells was evaluated at 3 h after treatment. The results of an MTT assay revealed that exposure of SH-SY5Y cells to Aβ1-42 for 3 h significantly reduced cell viability in a concentration-dependent manner ( Figure 3A). Based on these findings, we decided to examine neurotoxicity after exposure to 5 μM Aβ1-42 in SH-SY5Y cells. Viability significantly decreased with the 5 μM Aβ1-42 exposure, and the decrease was significantly recovered by treatment with 10 μM FA (p = 0.0135 vs. 5 μM Aβ1-42), and the combination of both (p < 0.0001 vs. 5 μM Aβ1-42). Moreover, the viability of cells treated with the combination of 1 uM Cur + 10 uM FA was increased compared to cells treated with 1 μM Cur alone (n = 6, Tukey, p = 0.0488).  As shown in Figures 3 and S1, to determine the neurotoxicity of Aβ 1-42 and to compare the effects of Cur, FA, and their combination on cells, the viability of SH-SY5Y cells was evaluated at 3 h after treatment. The results of an MTT assay revealed that exposure of SH-SY5Y cells to Aβ 1-42 for 3 h significantly reduced cell viability in a concentration-dependent manner ( Figure 3A). Based on these findings, we decided to examine neurotoxicity after exposure to 5 µM Aβ 1-42 in SH-SY5Y cells. Viability significantly decreased with the 5 µM Aβ 1-42 exposure, and the decrease was significantly recovered by treatment with 10 µM FA (p = 0.0135 vs. 5 µM Aβ 1-42 ), and the combination of both (p < 0.0001 vs. 5 µM Aβ 1-42 ). Moreover, the viability of cells treated with the combination of 1 uM Cur + 10 uM FA was increased compared to cells treated with 1 µM Cur alone (n = 6, Tukey, p = 0.0488).

Oxidative Stress
Since Aβ1-42 induces oxidative stress, oxidative stress is associated with Aβ1-42 [25]. It has been suggested that oxidative stress plays an important role in the pathogenesis of AD because increased oxidative stress contributes to cell membrane damage and cell death. Then, the protective effects of Cur, FA and the combination of both compounds on Aβ1-42-induced oxidative stress were investigated. Figure 5 shows the levels of ROS production in SH-SY5Y cells incubated with 5 μM Aβ1-42 for 30 min in the presence of Cur, FA, or the combination of both. ROS production significantly increased in SH-SY5Y cells exposed to Aβ1-42 (5 µM) compared with control cells. However, the increase due to Aβ1-42 exposure was significantly suppressed after 30 min of treatment with 1 µM Cur (p = 0.0466 vs. 5 μM Aβ1-42), 10 µM FA (p = 0.0003 vs. 5 μM Aβ1-42), or the combination of both (p < 0.0001 vs. 5 μM Aβ1-42). In particular, treatment with the combination of 1 µM Cur and 10 µM FA significantly decreased ROS production, compared to treatment with 1 μM Cur alone (Tukey, p = 0.0499) ( Figure 5A). In addition, the combination treatment of 5 μM Cur and 10 μM FA significantly reduced ROS production, compared to 5 μM Cur single treatment (Tukey, p = 0.0057) ( Figure S2A). The images obtained with a fluorescence microscope are shown in Figure Figure 5 shows the levels of ROS production in SH-SY5Y cells incubated with 5 µM Aβ 1-42 for 30 min in the presence of Cur, FA, or the combination of both. ROS production significantly increased in SH-SY5Y cells exposed to Aβ 1-42 (5 µM) compared with control cells. However, the increase due to Aβ 1-42 exposure was significantly suppressed after 30 min of treatment with 1 µM Cur (p = 0.0466 vs. 5 µM Aβ 1-42 ), 10 µM FA (p = 0.0003 vs. 5 µM Aβ 1-42 ), or the combination of both (p < 0.0001 vs. 5 µM Aβ 1-42 ). In particular, treatment with the combination of 1 µM Cur and 10 µM FA significantly decreased ROS production, compared to treatment with 1 µM Cur alone (Tukey, p = 0.0499) ( Figure 5A). In addition, the combination treatment of 5 µM Cur and 10 µM FA significantly reduced ROS production, compared to 5 µM Cur single treatment (Tukey, p = 0.0057) ( Figure S2A). The images obtained with a fluorescence microscope are shown in Figure 5B  Fluorescence intensity x 10 5 /µ g protein  Figure 6 shows the levels of mitochondrial ROS and Mn-SOD production in SH-SY5Y cells incubated with 5 µM Aβ 1-42 in the presence of Cur, FA, or combinations of both. As shown in Figure 6A, levels of mitochondrial ROS production were significantly increased by 5 µM Aβ 1-42 exposure and significantly suppressed by treatment with FA (p = 0.0019 vs. 5 µM Aβ 1-42 ) or combination of Cur + FA (p < 0.0001 vs. 5 µM Aβ 1-42 ). Combination treatment with Cur and FA significantly reduced mitochondrial ROS production compared to treatment with Cur alone (Tukey, p = 0.001) ( Figure 6A). Figure 6 shows the levels of mitochondrial ROS and Mn-SOD production in SH-SY5Y cells incubated with 5 μM Aβ1-42 in the presence of Cur, FA, or combinations of both. As shown in Figure 6A, levels of mitochondrial ROS production were significantly increased by 5 μM Aβ1-42 exposure and significantly suppressed by treatment with FA (p = 0.0019 vs. 5 μM Aβ1-42) or combination of Cur + FA (p < 0.0001 vs. 5 μM Aβ1-42). Combination treatment with Cur and FA significantly reduced mitochondrial ROS production compared to treatment with Cur alone (Tukey, p = 0.001) ( Figure 6A).

Mitochondrial ROS Production and Manganese Superoxide Dismutase (Mn-SOD) Levels
Superoxide dismutase has the role of protecting cells from ROS by dismutation of superoxide radicals to molecular oxygen and hydrogen peroxide. As shown in Figure 6B, Mn-SOD levels were significantly decreased in SH-SY5Y cells after Aβ1-42 exposure, which was reversed by treatment with 10 μM FA (p = 0.0197 vs. 5 μM Aβ1-42), but had no effect on Cur + FA treatment. Moreover, there was no marked effect on Mn-SOD levels in treated cells with their combination compared to Cur or FA alone.

Effects of Cur or FA and Their Combination on Aβ1-42-Induced Disruption of Membrane Integrity
Aβ1-42 is thought to bind directly to membrane lipids, damage the phospholipid bilayer structure, and invade cells [26,27]. In this study, changes in cell membrane fluidity and cell membrane phospholipid peroxidation due to Aβ exposure were investigated. Superoxide dismutase has the role of protecting cells from ROS by dismutation of superoxide radicals to molecular oxygen and hydrogen peroxide. As shown in Figure 6B, Mn-SOD levels were significantly decreased in SH-SY5Y cells after Aβ 1-42 exposure, which was reversed by treatment with 10 µM FA (p = 0.0197 vs. 5 µM Aβ 1-42 ), but had no effect on Cur + FA treatment. Moreover, there was no marked effect on Mn-SOD levels in treated cells with their combination compared to Cur or FA alone.

Effects of Cur or FA and Their Combination on Aβ 1-42 -Induced Disruption of Membrane Integrity
Aβ 1-42 is thought to bind directly to membrane lipids, damage the phospholipid bilayer structure, and invade cells [26,27]. In this study, changes in cell membrane fluidity and cell membrane phospholipid peroxidation due to Aβ exposure were investigated.

Fluidity of the Cell Membrane
The fluidity of cell membranes was significantly reduced a and the reduction was significantly suppressed by treatment wi Aβ1-42), Cur (p = 0.00348 vs. 5 μM Aβ1-42), or Cur + FA (p < 0.00 min. Compared with 1 μM Cur alone, combination treatment w vs. 1 μM Cur, Tukey) after exposure to Aβ1-42 resulted in a signif brane fluidity (Figure 7).

Phospholipid Peroxidation in the Cell Membrane
As shown in Figure 8, the phospholipid peroxidation in ce cantly increased by 5 μM Aβ1-42 exposure, and the increase wa by treatment with FA (p = 0.0012 vs. 5 μM Aβ1-42) or combinati vs. 5 μM Aβ1-42) for 30 min. The combination of Cur and FA si pholipid peroxidation levels compared to treatment with Cur

Phospholipid Peroxidation in the Cell Membrane
As shown in Figure 8, the phospholipid peroxidation in cell membranes was significantly increased by 5 µM Aβ 1-42 exposure, and the increase was significantly suppressed by treatment with FA (p = 0.0012 vs. 5 µM Aβ 1-42 ) or combination of Cur + FA (p = 0.0009 vs. 5 µM Aβ 1-42 ) for 30 min. The combination of Cur and FA significantly reduced phospholipid peroxidation levels compared to treatment with Cur alone (p = 0.0009, Tukey). Moreover, the combination of 5 µM Cur + 10 µM FA led to a significant reduction, compared to treatment with 5 µM Cur alone (p = 0.0006, Tukey) ( Figure S3B).

Changes in Intracellular Calcium ([Ca 2+ ]i) Following Treatment Combination of Both
As shown in Figure 9 anf Figure S4

Discussion
In AD, Aβ abnormalities precede the onset of cognitive d mately 25 years [28]. Therefore, it is believed that starting treatm before the onset of the disease, rather than after the onset of cogni the onset of cognitive dysfunction. Furthermore, currently appro AChE inhibitors and NMDA receptor antagonists, constitute o ments with minor effects and no impact on long-term disease prog prophylactic approaches are, however, considered essential for re treatment. One such practical long-term approach is the intake of supplements and natural products such as fruits, vegetables, an daily intake of specific agents might be difficult to maintain whe plements. Therefore, the combined use of low doses of the targeted provide synergistic effects and help individuals overcome these d In the current study, we selected Cur and FA as active ingre of Aβ abnormalities. Because each has a different mechanism of a may directly improve the disease state of multiplex AD. The mec on AD has been reported to involve the inhibition of Aβ aggregatio

Discussion
In AD, Aβ abnormalities precede the onset of cognitive dysfunction by approximately 25 years [28]. Therefore, it is believed that starting treatment from an early stage before the onset of the disease, rather than after the onset of cognitive dysfunction, delays the onset of cognitive dysfunction. Furthermore, currently approved AD drugs, such as AChE inhibitors and NMDA receptor antagonists, constitute only symptomatic treatments with minor effects and no impact on long-term disease progression [29]. Long-term prophylactic approaches are, however, considered essential for realistic measures for AD treatment. One such practical long-term approach is the intake of readily available dietary supplements and natural products such as fruits, vegetables, and seeds. However, high daily intake of specific agents might be difficult to maintain when taken as dietary supplements. Therefore, the combined use of low doses of the targeted active ingredients may provide synergistic effects and help individuals overcome these difficulties.
In the current study, we selected Cur and FA as active ingredients for the treatment of Aβ abnormalities. Because each has a different mechanism of action, their combination may directly improve the disease state of multiplex AD. The mechanism of action of Cur on AD has been reported to involve the inhibition of Aβ aggregation, an increase in BDNF, and a decrease in tau phosphorylation [10,30,31]. On the other hand, research has shown that FA inhibits Aβ production via downregulation of APP and β-secretase [32], has potent antioxidant properties, and protects neurons from Aβ-induced neurotoxicity mainly although we previously confirmed FA also has anti-amyloidogenic effect for Aβ [21]. We hypothesized that FA and Cur might complement each other based on such reports.
As one of the features of AD pathology, Aβ accumulation in the brain causes conformational changes in peptides, forming oligomers and fibrils that deposit on plaques [5,30]. It has recently been proposed that the aggregation mechanism of Aβ peptide is that soluble Aβ monomers self-associate via conformational changes to form β-sheet-rich oligomers, and that the monomers further bind to the oligomers and elongate to form Aβ fibrils [5,33]. Here, the effect of Cur, FA and their combination on the aggregation of Aβ 1-40 and Aβ 1-42 peptides was monitored through temporal changes in fibrillar β-sheet content using ThT assays. Cur inhibited the aggregation of Aβ 1-42 at low concentrations, and the inhibitory effect was dose-dependent ( Figure 2B). Both Aβ 1-40 and Aβ 1-42 were used at 25 µM in the ThT assay. Cur showed an inhibitory effect on Aβ aggregation at a low concentration of 1 µM, indicating that Cur inhibited aggregation even at a concentration of 1/25 of that of Aβ. Cur has been reported to inhibit Aβ aggregation including oligomerization in vitro and in vivo [9,10]. Furthermore, Cur binds directly to Aβ and thereby inhibits Aβ aggregation [10]. An NMR analysis showed that Cur interacts with amino acid residues number 12 and 17-21 of Aβ 1-42 [34], suggesting that it has an inhibitory effect on Aβ fibril elongation. The plant polyphenols myricetin, morin, and EGCG were previously reported to have an inhibitory effect on Aβ aggregation, including oligomerization. Myricetin and morin inhibit β-sheet-rich oligomer formation from soluble Aβ monomers [35]. EGCG inhibits Aβ fibril formation by promoting non-toxic oligomer formation ("off-pathway" aggregation) and inhibits Aβ fibril formation [36]. Several studies have reported that the phenolic hydroxyl group of polyphenol compounds that bind to histidine is required for anti-Aβ aggregation [37]. Cur has been shown to prevent the peptide-peptide interaction between Phe15 and His18, which is essential for Aβ aggregation [38]. Furthermore, quinones generated from phenolic hydroxyl groups react with the Lys side chains of proteins [39]. These polyphenolic compounds react with Lys28, which is essential for Aβ aggregation and may contribute to the inhibition of Aβ aggregation [40]. Moreover, quinones produced from phenolic hydroxyl groups may react with the Lys28 side chains of Aβ peptide and contribute to the inhibition of Aβ aggregation. On the other hand, in this experiment, FA had no significant effect on Aβ aggregation, but the combination of Cur and FA led to significant inhibition of aggregation of both Aβ 1-40 and Aβ 1-42 compared to Aβ alone at endpoint (Figure 2A,B). The combination of Cur and FA may result in a coordination with each other, such that Cur prevented the peptide-peptide interaction of Aβ and FA reacted with the Lys residue of Aβ.
Oxidative stress is normally regulated by the antioxidant defense system. However, in many diseases, that system is disrupted, and oxidative stress is known to play an essential role in pathogenesis. In particular, it leads to the accumulation of oxidative damage in AD, is associated with age-related neurodegenerative diseases, and represents the most common cause of dementia in the elderly. Aβ induces oxidative stress in vivo and in vitro and is considered an early event in AD as it contributes to membrane damage and cell death [41]. Furthermore, the extensive oxidative damage observed in brain regions with mild cognitive impairment (MCI) also suggests that oxidative stress may be an early event in the progression from normal aging to AD [42]. However, how Aβ causes oxidative stress is currently unknown. In the present study, Aβ 1-42 increased ROS, mitochondrial-ROS, and the peroxidation of plasma membrane phospholipids in SH-SY5Y ( Figures 5, 6A and 8). Since Aβ 1-42 was exposed extracellularly, it is possible that Aβ 1-42 first contacted the cell membrane and induced phospholipid peroxidation. Many studies have shown that Aβ as an oligomer inserts into the plasma membrane bilayer and initiates lipid peroxidation [43]. Aβ-induced lipid peroxidation then promotes Ca 2+ influx into neurons, increases toxicity, and facilitates apoptosis [44]. Considering the report by Nakayama et al., demonstrating an increase in fibrils in Aβ after 3 h of incubation, it is possible that the Aβ 1-42 used in the current experiment contained more fibrils than the oligomers [45] and that even the fibril-rich Aβ 1-42 induced peroxidation of plasma membrane phospholipids and increased mitochondrial-ROS and [Ca 2+ ] i. (Figures 6A, 8 and 9). The increase in cytoplasmic Ca 2+ with Aβ 1-42 exposure has been shown to cause disruption of mitochondrial homeostasis, leading to the production of ROS [46]. Mitochondrial dysfunction and oxidative stress are interacting processes. Increased oxidative stress leads to decreased glucose metabolism and ATP synthesis in the brain. Thus, oxidative stress is closely associated with Aβ neurotoxicity and plays an important role in the pathological mechanisms underlying AD. Both Cur and FA are polyphenols with antioxidant activity, and especially FA exhibits high antioxidant activity. In the current study, combined treatment with Cur and FA induced stronger antioxidant effects against Aβ 1-42 -induced oxidative stress, such as lower ROS levels, lower plasma membrane phospholipid peroxidation, and lower mitochondrial ROS levels, compared to Cur and FA treatment alone (Table 1, Figures 5, 6A and 8). FA also has antioxidant properties due to its phenolic hydroxyl group, while the hydroxy and phenoxy groups donate electrons to scavenge free radicals. In in vivo experiments, FA has been shown to protect against Aβ 1-42 -induced oxidative stress and neurotoxicity in primary cortical neurons in rats [47]. In vitro studies have shown that FA inhibits Aβ 1-42 -induced cell death and apoptosis in LAN5 neuroblastoma cells [48], indicating that it may be beneficial in the prevention and treatment of AD. Similarly, in the present experiments, FA treatment alone significantly inhibited Aβ 1-42 -induced oxidative stress. The o-methoxyphenyl group and methylene hydrogen in Cur contribute to the compound's antioxidant activity, by donating electrons/hydrogen atoms to ROS and neutralizing reactive oxygen intermediates [49]. The significant antioxidant effect of Cur + FA on Aβ-induced oxidative stress may thus be the result of a complex mechanism. In the present experiment, Aβ 1-42 exposure reduced mitochondria-localized Mn-SOD and treatment with Cur and FA could restore the expression of Mn-SOD protein ( Figure 6B). A study in vitro reported that Aβ directly interfered with mitochondrial respiration [50]. SOD is a reactive oxygen species scavenger, acting on ROS generation and protecting cells from cellular damage due to oxidative stress. However, combined treatment with Cur and FA showed no significant effect on Mn-SOD levels ( Figure 6B). Furthermore, 5 µM Cur treatment eliminated the difference compared to Aβ 1-42 exposure, and combined treatment with 5 µM Cur + 10 µM FA significantly reduced Mn-SOD level ( Figure S2C). In human cancer cells, Cur has been shown to convert its antioxidant effect into an oxidant-promoting effect [49], indicating that Cur exhibits antioxidant activity in normal cells under stressful conditions and waits for the biphasic activity of the oxidant-promoting effect in cancer cells. In our cell viability experiments at 24 h incubation, the combined treatment with 5 µM Cur + 10 µM FA showed no significant effect on viability compared to Aβ 1-42 exposure (5 µM Cur: 53.20 ± 0.62 vs 5 µM Aβ 1-42 : 50.67 ± 1.94, n = 6, Tukey). Decreased Mn-SOD levels with 5 µM Cur + 10 µM FA combination treatment may be one of the reasons for the decreased cell viability. Prolonged exposure to Cur + FA combinations at high concentrations may reduce cell viability and should be noted.
In the present experiments, Aβ 1-42 exposure induced the disruption of membrane integrity and increased [Ca 2+ ] i (Figures 7-9). Alterations in the functional integrity of neuronal membranes in AD may result from interactions between Aβ and the membrane. Aβ oligomers cause membrane permeation through several hypothetical mechanisms such as ion channels in cell membranes and transmembrane oligomer pore structure formation [51]. Since the Aβ used here is presumed to have a higher fibril content than the oligomers, the Aβ-induced disruption of membrane integrity shown in Figures 7 and 8 may be due to Aβ-induced oxidative stress rather than Aβ-induced formation of ion channels. The generated ROS, especially the most reactive hydroxyl radicals, cause oxidative damage to both the Aβ peptide itself and the surrounding proteins and lipids. Lipid peroxidation products of cell membranes then bind to several membrane proteins, thereby altering their protein structure and function and resulting in changes in neurotoxicity [25]. Moreover, as shown in Figure 7, Aβ 1-42 decreased membrane fluidity. Reduced membrane fluidity by Aβ accelerates the amyloidogenic processing of APP [52].
Conversely, increased membrane fluidity shifts APP cleavage processing by α-secretase to non-amyloidogenic [53]. In vivo experiments have shown that Aβ administration reduces the membrane fluidity of synaptosomes isolated from frontal and hypothalamic neurons of 3-month-old mice [54]. In the current study, we observed that treatment with Cur, FA and Cur + FA restored membrane fluidity reduced by Aβ   (Figure 7). FA (100 µmol/L) reduced cholesterol levels in erythrocytes as well as lipid peroxidation when incubated with human erythrocytes for 24 h [55]. Cholesterol is an essential component of cell membranes, and higher membrane cholesterol levels reduce cell membrane fluidity. Depletion of cell membrane cholesterol, in contrast, results in increased membrane fluidity. One cause of the increase in cell membrane fluidity induced by FA may be that FA reduces cell membrane cholesterol levels. Cholesterol content in mammalian phospholipid bilayers is high, ranging from 20-30% in most cells and 40-50% in erythrocytes. Cur increases the membrane fluidity of liposome membranes containing less than 20% cholesterol and diffuses into the membrane, but stiffens the liposome membranes containing 40% cholesterol [56]; Cur may thus act directly on cholesterol in the liposome membrane. In the present experiments, the combined treatment with Cur + FA significantly increased fluidity compared to Cur and FA treatment alone (Figure 7). This indicates that Cur acted directly on cell membrane cholesterol in the SH-SY5Y cells, that FA reduced cell membrane cholesterol levels and that the combination of Cur + FA increased membrane fluidity through both actions.
One of the challenges of developing drugs for AD is to design drugs that improve symptoms but have few side effects. Safety is an even more critical issue in combination treatments than in single treatments, and side effects need to be considered. The acute oral lethal dose with 50% survival (LD 50) is >2000 mg/kg for Cur [57] and FA [58] when administered to mice, suggesting that both compounds are safe. Pharmacokinetics studies of oral administration of 10 g of Cur to healthy human volunteers showed a Cmax of 2.3 µg/mL (7 µM) [59]. The concentration of Cur (1 or 5 µM) used in this experiment seems to be suitable.
Considering the safety of Cur, in vivo studies indicated that Cur is poorly absorbed from the intestine when orally administered to rats in a single dose (2 g), and that plasma concentrations were below 5 µg/mL. In brief, the tissue concentration of Cur does not lead to beneficial or detrimental effects, due to insufficient absorption via the oral route, which makes it safe for oral administration. On the other hand, FA is absorbed from the stomach and small intestine, and unrestricted FA has been found in the human plasma only 10 min after oral administration of sodium ferulate: FA thus has good bioavailability [60]. Therefore, the safety of long-term Cur and FA combination treatment should be investigated in vivo.
We find that combination treatment of Cur and FA exerts a cytoprotective effect on Aβ-induced cytotoxic effects, through multiple mechanisms. These mechanisms include the suppression of Aβ aggregation and antioxidant effects, as compared to single treatment with either Cur or FA alone ( Table 1). The protective effects of the combination treatment we observed were complementary and cooperative. These findings suggest that the combination of Cur and FA may provide an effective and superior strategy for the prevention and therapeutics of AD in humans.

Drugs and Reagents
Human amyloid β-protein (Aβ, Human, 1-42) was purchased from Peptide Institute (Osaka, Japan). DMEM Ham's F-12 medium and all-trans retinoic acid (ATRA) were purchased from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan). Penicillin G sodium, streptomycin sulfate, amphotericin B, fetal bovine serum (FBS), Cur, and FA were obtained from Thermo Fisher Scientific K.K. (Waltham, MA, USA). Cur and FA were dissolved in dimethyl sulfoxide (DMSO) and then disbanded in medium to achieve a final concentration of DMSO to 0.1%. The other chemicals used in this experiment were the purest commercially available.

Cell Culture and Drug Treatment
SH-SY5Y cells (human neuroblastoma, EC-94030304) were obtained from the European Collection of Authenticated Cell Cultures (London, UK). SH-SY5Y cells were cultured in DMEM Ham's F-12 containing 10% FBS, penicillin G sodium, streptomycin sulfate, and amphotericin B, and maintained in a humid atmosphere of 5% CO 2 and 95% air at 37 • C. SH-SY5Y cells were treated with 10 µM ATRA for 5 days to differentiate. Aβ 1-42 was dissolved in DMSO and incubated at 37 • C for 24 h for self-aggregation, and then prepared to 5 µM in a DMEM Ham's F-12 medium without fetal bovine serum (FBS). The differentiated SH-SY5Y cells were then treated with 5 µM Aβ 1-42 containing Cur (1, 5 µM), FA (10 µM) or the combination of both lysed in DMEM / Ham's F-12 medium. As a control, cells cultured in a medium containing 0.1% DMSO were used. All treatments were performed aseptically.

Cell Viability Assay
An MTT assay was applied to evaluate the effect of Cur, FA and the combination of both compounds on the viability of SH-SY5Y cells exposed to Aβ  . MTT assay is based on the formation of blue formazan metabolized from colorless MTT by mitochondrial dehydrogenases, which are active only in live cells. The Cell Proliferation Kit I (11465007001, Roche, Mannheim, Germany) was used according to the manufacturer's instructions. The differentiated SH-SY5Y cell of 1.0 × 10 5 cells/mL were seeded into 96-well collagen-coated plates and incubated at 37 • C for 24 h. First, to assess a suitable concentration of Aβ 1-42 to induce cytotoxicity in SH-SY5Y cells, preliminary experiments were performed in which SH-SY5Y cells were exposed to Aβ 1-42 (1, 5 and 10 µM) for 3 h. As shown in Figure 3A, the appropriate concentration of Aβ 1-42 to induce cytotoxicity was 5 µM. Next, to investigate the protective effects of Cur, FA and the combination of both compounds on Aβ 1-42 -induced cytotoxicity, SH-SY5Y cells were treated with Aβ 1-42 + Cur (1, 5 µM), Aβ 1-42 + FA (10 µM), or Aβ 1-42 + Cur + FA for 3 h. After incubation, a MTT assay was performed and the results were measured at 540 nm using a microplate reader Spectra Max i3 (Molecular Devices Co., San Jose, CA, USA).

Calcein-AM and EthD-1 (Live/Dead) Cell Assay
Live cells and dead cells were also observed by calcein-AM and EthD-1 costaining. SH-SY5Y cells were seeded at 1.0 × 10 6 cells/mL in 96-well collagen-coated plates and incubated at 37 • C for 24 h, then exposed to Aβ 1-42 or treated with Aβ 1-42 + Cur, Aβ 1-42 + FA or Aβ 1-42 + Cur + FA for 3 h. The treated cells were stained with 2 µM calcein-AM and 10 µM EthD-1 (Thermo Fisher Scientific K.K, Waltham, MA, USA). The green fluorescent calcein, hydrolyzed by ubiquitous intracellular esterase in the cells, depends on the number of live cells, while EthD-1 only enters cells with damaged membranes, binds to nucleic acids, and emits bright red fluorescence proportional to the number of dead cells. Using Spectra Max i3 (Molecular Devices), the green fluorescence intensity was measured at Ex: 495 nm and Em: 530 nm, and the red fluorescence intensity at Ex: 495 nm and Em: 645 nm. The morphology of individual cells was also evaluated by observation with a fluorescence microscope (BZX800; Keyence, Osaka, Japan).

Detection of Manganese-Superoxide Dismutase (Mn-SOD)
SH-SY5Y cells were incubated at 37 • C for 24 h, after which cells were exposed to Aβ 1-42 and treated with Cur for 24 h. The mitochondrial SOD isozyme content in cell lysates was determined and measured by ELISA using a monoclonal antibody (Human SOD2 ELISA Kit; ab178012, Abcam, Cambridge, UK). Absorbance was measured at 450 nm using Spectra Max i3 (Molecular Devices). The protein concentration of cell lysate was determined using the protein assay dye reagent (Bio-Rad Laboratories, Inc., Hercules, CA, USA). The dynamic properties of cell membranes are important because they are associated with various pathological syndromes associated with membrane fluidity. Damage to the membrane of neurons by toxic Aβ  has been hypothesized to be a major event of neurotoxicity in AD. To understand the interaction of Aβ 1-42 with the lipid bilayer, we measured cell membrane fluidity. The membrane fluidity of SH-SY5Y cells was measured using the lipophilic pyrene probe pyrene decanoate (PDA) of the Membrane Fluidity Kit (ab189819, Marker Gene Technologies, Inc., Eugene, OR, USA). SH-SY5Y cells at 1.0 × 10 6 cells/mL were exposed to 5 µM Aβ 1-42 or treated with Aβ 1-42 + Cur, Aβ 1-42 + FA, or Aβ 1-42 + Cur + FA for 30 min. The treated cells were stained with PDA, which causes excimer formation by spatial interaction, and membrane fluidity was measured according to a previously described method [40]. The ratio of monomer (Em: 372 nm) to excimer (Em: 470 nm) fluorescence was measured with a Spectra Max i3 (Molecular Devices).

Assay of Phospholipid Peroxidation in Cell Membranes
To detect the peroxidation of phospholipids in the membranes of cells, SH-SY5Y cells (1.0 × 10 6 cells/mL) were stained with 5 µM diphenyl-1-pyrenylphosphine (DPPP; Thermo Fisher Scientific K.K, Waltham, MA, USA) in DMSO at 37 • C for 10 min, and then phospholipid peroxidation was measured according to a previously described method [61]. Briefly, the stained cells were exposed to Aβ 1-42 or treated with Aβ 1-42 + Cur, Aβ 1-42 + FA, or Aβ 1-42 + Cur + FA for 30 min. The fluorescence intensity of DPPP oxide was monitored using Spectra Max i3 (Molecular Devices) at Ex: 351 nm and Em: 380 nm. DPPP is known to react quantitatively with hydroperoxides to produce strong fluorescent DPPP oxides.

Detect of Changes in [Ca 2+ ] i
To observe changes in [Ca 2+ ] i levels in SH-SY5Y cells, we used the FLIPR Calcium 5 Assay Kit (R8185, Molecular Devices). In brief, the differentiated SH-SY5Y cells were loaded with FLIPR reagent containing 20 mM HEPES and 1 × Hank's Balanced Salt solution (pH 7.4) for 60 min at 37 • C in the presence of DMEM Ham's F-12 medium. Then, 5 µM Aβ 1-42 , 1 µM Cur, 10 µM FA, or 1 µM Cur + 10 µM FA were added 20 s after the start of measurement. Change in [Ca 2+ ] i was monitored at an excitation wavelength of 485 nm and an emission wavelength of 525 nm at 37 • C for 300 s at 3-s intervals using Spectra Max i3 (Molecular Devices). The fluorescence intensity at onset was expressed as 100%.

Statistical Analysis
Each measurement was performed in triplicate. Results were expressed as mean + S.E.M. The effects of Cur, FA or Cur + FA were compared with SH-SY5Y cells exposed to 5 µM Aβ 1-42 , which was also included in other reagents, using analysis of variance (ANOVA) followed by Tukey or Dunnett's post hoc test. A value of p < 0.05 was considered statistically significant for all tests.