EK100 and Antrodin C Improve Brain Amyloid Pathology in APP/PS1 Transgenic Mice by Promoting Microglial and Perivascular Clearance Pathways

Alzheimer’s disease (AD) is characterized by the deposition of β-amyloid peptide (Aβ). There are currently no drugs that can successfully treat this disease. This study first explored the anti-inflammatory activity of seven components isolated from Antrodia cinnamonmea in BV2 cells and selected EK100 and antrodin C for in vivo research. APPswe/PS1dE9 mice were treated with EK100 and antrodin C for one month to evaluate the effect of these reagents on AD-like pathology by nesting behavior, immunohistochemistry, and immunoblotting. Ergosterol and ibuprofen were used as control. EK100 and antrodin C improved the nesting behavior of mice, reduced the number and burden of amyloid plaques, reduced the activation of glial cells, and promoted the perivascular deposition of Aβ in the brain of mice. EK100 and antrodin C are significantly different in activating astrocytes, regulating microglia morphology, and promoting plaque-associated microglia to express oxidative enzymes. In contrast, the effects of ibuprofen and ergosterol are relatively small. In addition, EK100 significantly improved hippocampal neurogenesis in APPswe/PS1dE9 mice. Our data indicate that EK100 and antrodin C reduce the pathology of AD by reducing amyloid deposits and promoting nesting behavior in APPswe/PS1dE9 mice through microglia and perivascular clearance, indicating that EK100 and antrodin C have the potential to be used in AD treatment.


Introduction
Alzheimer's disease (AD) is characterized by a progressive decline in cognitive abilities that can ultimately affect daily living activities. Extracellular deposition of β-amyloid (Aβ) plaques and intracellular neurofibrillary tangles are considered the main pathological Eburicoic acid and dehydroeburicoic acid-treated mice reduced hyperglycemia, hypertriglyceridemia, hyperinsulinemia, hyperleptinemia, and hypercholesterolemia induced by a high-fat diet [23]. Sulfuric acid showed protective effects on type 1 diabetes and hyperlipidemia in diabetic mice induced by streptozotocin [24]. However, the anti-Alzheimer's disease effect of these compounds has never been studied. APP/PS1 transgenic mice co-expressing Swedish mutant human APP695 and mutant human presenilin 1 (PS1) (in which exon 9 is deleted) [25] exhibit pathological and behavioral changes similar to AD, including amyloid in the brain accumulation of plaques, degeneration of the cholinergic system, and impaired exploratory behavior and spatial memory [26]. As early as 3 to 5 months of age, APP/PS1 mice have increased Aβ production and plaque formation [27], and impairments in spatial learning and memory are also observed at 6 months of age [28,29]. In order to verify the effects of the four compounds on behavioral disorders in APP/PS1 mice, we focused on species-specific nesting activities, because it is multi-brain-dependent spontaneous [30], which has been considered similar to activities of daily living (ADL) skills [31]. Clinically, ADL disorder is a pathological manifestation of AD [32], and this pathological manifestation also appears in APP/PS1 mice [29]. In addition, it was found that the hippocampal neurogenesis of APP/PS1 mice was damaged at 3 to 6 months of age [33]. Our previous studies have shown that antiinflammatory effects can promote hippocampal neurogenesis [29], so we hypothesized that the decline in microglia activation may subsequently promote hippocampal neurogenesis.
The purpose of this study is to study the effects of EK100, antrodin C, ergosterol, and ibuprofen on AD-related pathology in APP/PS1 transgenic mice.

EK100 and Antrodin C Confer Anti-Inflammatory Effects on BV2 Microglia
The structure of EK100 and antrodin C as well as their related compounds used in this study are shown in Figure 1.

EK100 and Antrodin C Reduce the Number and Burden of Amyloid Plaques in the Brains of APP/PS1 Mice
It is known that plaques can be clearly observed in the brains of APP/PS1 mice at 6 months of age [29]. Therefore, the effect of the four compounds on the number and size distribution of plaque were detected by immunostaining with AB10 antibody. The plaque number and burden were calculated using MetaMorph software. The plaque number and burden were significantly reduced after 30-day administration of EK100 or antrodin C. Compared with vehicle-treated mice, EK100 and antrodin C decreased plaque number by 45 Figure S1), and they decreased plaque burden by 37.5% (1.31 ± 0.24 vs. 2.13 ± 0.19, p < 0.05, n = 5) and 53.9% (0.98 ± 0.12 vs. 2.13 ± 0.19, p < 0.01, n = 5), respectively ( Figure 3A,C). In contrast, ergosterol and ibuprofen did not have any significant effect on the number of plaques, and ibuprofen even increased the plaque load by 31.6% (2.80 ± 0.06 vs. 2.13 ± 0.19, p < 0.05, n = 5).
Since the decrease in the number of plaques may be due to a decrease in Aβ levels in the brain, we subsequently measured Aβ levels in the hippocampus. However, after all four treatments, there was no significant change in Aβ levels in the hippocampus (Supplementary Figure S2A). In contrast, serum Aβ1-42 levels significantly decreased after the ergosterol and ibuprofen treatment, while serum levels of Aβ1-40 significantly increased after antrodin C treatment (Supplementary Figure S2B).

EK100 and Antrodin C Promote Aβ Perivascular Deposition in the Brain of APP/PS1 Mice
Since Aβ in the ISF of the brain can be removed from the brain through the glymphatic perivenous drainage pathway and/or IPAD pathway [34], perivascular Aβ deposition is detected in both the cortex and hippocampus. We found that the distribution of Aβ deposits changed from amyloid plaques to Aβ deposits around blood vessels after treatment with EK100 and antrodin C ( Figure 4). The calculation of Aβ deposition in the perivascular area is expressed as a percentage of the total deposition ( Figure 4B). The results showed that EK100 and antrodin C increased the perivascular Aβ deposition in the cortex by 22.86% (63.97 ± 5.38 vs. 41.12 ± 3.71, p < 0.01, n = 5 vs. 6) and 19.89% (61.00 ± 0.84 vs. 41.12 ± 3.713, p < 0.01, n = 5 vs. 6), respectively. These changes were found in both capillaries and arteries (Supplementary Figure S1C). Similarly, EK100 and antrodin C increase the perivascular Aβ deposition in the hippocampus by 31.67% (76.48 ± 13.05 vs. 44.80 ± 6.52, p < 0.05, n = 5 vs. 6) and 28.21% (73.01 ± 4.84 vs. 44.80 ± 6.52, p < 0.05, n = 5 vs. 6), respectively.

EK100 and Antrodin C Reduce the Number of Glial Cluster and Glial Activation in the Brain of APP/PS1 Mice
M2-microglia phagocytosis is another Aβ clearance pathway. Therefore, changes in the phenotype of microglia may also help reduce amyloid plaques. After 30 days of treatment with EK100 and antrodin C, the number of clusters formed by PAM decrease significantly ( Figure 3A,C). Compared with vehicle-treated mice, EK100 treatment reduced the number of clusters formed by PAM by 35.5 ± 9.6% (37.40 ± 2.50 vs. 57.83 ± 2.07, p < 0.001, n = 5). The less effective is that antrodin C treatment reduces the number of clusters containing PAM by 34.2 ± 18.5% (41.40 ± 5.66 vs. 57.83 ± 2.07, p < 0.05, n = 5). In contrast, ergosterol and ibuprofen had no significant effect on the number of clusters formed by PAM.
On the other hand, compared with vehicle-treated mice, EK100 treatment reduced number of clusters formed by PAA by 42.0 ± 7.9% (34.80 ± 2.11 vs. 59.67 ± 1.84, p < 0.0001, n = 5) ( Figure 3A,D). The less effective is that antrodin C treatment reduces the number of clusters containing PAA by 34.4 ± 20.1% (41.20 ± 4.62 vs. 59.67 ± 1.84, p < 0.01, n = 5). In contrast, ergosterol and ibuprofen had no significant effect on the number of clusters formed by PAA and PAM.
In order to determine the changes in microglia associated with activated plaques after drug treatment, a scatter plot of the fluorescence intensity of small and medium plaques (<21 pixels) and Iba-1 was drawn (Supplementary Figure S3). Linear regression analysis showed that the slope of the regression equation of EK100, antrodin C, and ibuprofen groups were significantly different from that of the vehicle group. Linear regression for the vehicle group is y = 0.51x + 0.76, R 2 = 0.06; for the EK100 group is y = 1.12x + 0.47, R 2 = 0.23 (p < 0.01 for different in slope); for the antrodin C group is y = 1.05x + 3.32, R 2 = 0.18 (p < 0.01 for different in slope); for the ergosterol group is y = 0.84x + 4.47, R 2 = 0.21 (p = 0.13 for different in slope); for the ibuprofen group is y = 0.89x + 0.95, R 2 = 0.21 (p < 0.01 for different in slope). This result indicates that EK-100, antrodin C, and ibuprofen, but not ergosterol, changed the relationship between plague and microglia activation.

EK100 and Antrodin C Reduce Non-Clustered Activation of Glia in the Hippocampus of APP/PS1 Mice
Previously, we found that compared with wild-type mice, APP/PS1 mice had a higher degree of astrocyte reactivity and microglia activation in areas unrelated to plaque [29]. Therefore, the immunoreactivity of the non-clustered glial cells in the hippocampus of each group was compared ( Figure 5). The results show that compared with wild-type mice, APP/PS1 mice have a higher degree of glial non-cluster activation. Compared with vehicletreated mice, EK100 and ibuprofen treatment reduced the IR of non-clustered astrocytes (NCA) in the hippocampal CA1 by 46.7% (9.49 ± 0.42 vs. 17.80 ± 1.98, p < 0.01, n = 5 vs. 6) and 47.8% (9.30 ± 0.50 vs. 17.80 ± 1.98, p < 0.05, n = 5 vs. 6), respectively ( Figure 5A,B). In contrast, antrodin C and ergosterol did not show a significant effect on the IR of NCA in the hippocampus. On the other hand, EK100, antrodin C and ibuprofen treatment reduced the IR of non-clustered microglia (NCM) in hippocampal CA1 by 50.7% (5.87 ± 0.61 vs. 11.90 ± 1.03, p < 0.001, n = 5 vs. 6), 54.7 ± 9.7% (5.14 ± 0.67 vs. 11.90 ± 1.03, p < 0.001, n = 5 vs. 6) and 38.1% (7.37 ± 1.37 vs. 11.90 ± 1.03, p < 0.05, n = 5 vs. 6), respectively ( Figure 5B). In contrast, ergosterol did not show a significant effect on the IR of NCM in the hippocampus. The similar changes were found in CA3 and DG areas. Microglia have different morphological responses to changes in brain physiology, from hyper-ramified form to amoeba form [35]. Since microglia fine-tune the function of neurons and glial through cell-to-cell crosstalk [36], the morphology of microglia can be used as an indicator of a variety of cell functions and dysfunctions in the brain. The skeletal analysis was used for regional analysis of multiple microglia in the region of interest. The results shown that EK100, ergosterol, and ibuprofen increase the number of branches, junctions, and endpoints of microglia ( Figure 5C,D). EK100, ergosterol, and ibuprofen treatment increased the branch number of microglia by 11.6% (5.56 ± 0.06 vs. 4.98 ± 0.09, p < 0.0001, n = 20), 18.3% (5.89 ± 0.12 ± 0.06 vs. 4.98 ± 0.09, p < 0.0001, n = 25 vs. 20), and 9.4% (5.45 ± 0.11 vs. 4.98 ± 0.09, p < 0.0001, n = 25 vs. 20), respectively. The similar changes were found in the intersection number and endpoint number of microglia.
Next, the combined immunostaining of Iba-1 and AB10 was used to check the microglia phagocytosis of Aβ. The results showed that antrodin C, ergosterol, and ibuprofen, significantly increased the accumulation of Aβ in microglia by 37 Figure 5E,F). On the contrary, EK100 did not alter the accumulation of Aβ in microglia.

Discussion
This study shows that two anti-inflammatory compounds, EK100 and antrodin C, selected from components isolated from A. cinnamomea mycelium, can reduce AD-like pathological changes in APP/PS1 mice in different ways when administered orally. BV2 cells activated by LPS were used to determine the anti-inflammatory effects of components isolated from A. cinnamomea mycelium (including CA-Et, anticin K, eburicoic acid, dehydroeburicoic acid, sulfurenic acid, dehydrosulfurenic acid, EK100, antrodin C, and ergosterol), and ibuprofen (an NSAID control). The results showed that the production of nitric oxide in BV2 cells activated by LPS was significantly inhibited by CA-Et, EK100, antrodin C, and ergosterol, but it was not affected by anticin K, eburicoic acid/dehydroeburicoic acid mixture, sulfurenic acid/dehydrosulfurenic acid mixture, and ibuprofen. Therefore, we examined the effects of EK100, antrodin C, ergosterol, and ibuprofen (an NSAID control) on the pathology of APP/PS1 transgenic mice, including changes in nesting behavior, plaque deposition, perivascular deposition, nerves glial activation, and the Nrf2/HO-1/NQO-1 signaling pathway. Moreover, the effect of EK100 on neurogenesis was also examined.
The results of the animal study are summarized in Table 2. In the behavior test, EK100, antrodin C, and ibuprofen, but not ergosterol, significantly improve the nesting ability of APP/PS1 mice. Among the effects related to Aβ deposition, EK100 and antrodin C can effectively reduce the number and load of plaques, promote Aβ deposition around blood vessels, and reduce the accumulation of PAA and PAM. Among the changes in PAM activation, EK100 significantly increased the expression of Iba-1 but did not change the ratios of HO-1/Iba-1 and NQO-1/Iba-1; ADC significantly increased Iba-1 and the HO-1/Iba-1 and NQO-1/Iba-1 ratios; ergosterol significantly increased the ratios of HO-1/Iba-1 and NQO-1/Iba-1 but did not increase the expression of Iba-1; and ibuprofen significantly increased the NQO-1/Iba-1 ratio and reduced the NQO-1/Iba-1 ratio without increasing the expression of Iba-1. There are no compounds that can modulate the expression of GFAP in PAA. In the activation of non-plaque-associated glial cells, EK100, antrodin C, and ibuprofen significantly increased the expression of Iba-1 in NCM, while antrodin C, ergosterol, and ibuprofen significantly increased the intracellular accumulation of Aβ in NCM. On the other hand, EK100, ergosterol, and ibuprofen significantly increased the ramification of NCM, while EK100 and ibuprofen can significantly reduce the expression of GFAP in NCA. In the response of non-plaque-related neurons, EK100, antrodin C, and ibuprofen significantly reduced the expression of Nrf2 in cortical neurons. Finally, EK100 has also been shown to significantly promote hippocampal neurogenesis.
In order to verify the effects of the four compounds on behavioral disorders in APP/PS1 mice, we focused on species-specific nesting activities, because it is multi-braindependent spontaneous [30], which has been considered similar to ADL skills [31]. Clinically, ADL disorder is a pathological manifestation of AD [32], and this pathological manifestation also appears in APP/PS1 mice [29]. In this study, we found that EK100, antrodin C, and ibuprofen can alleviate the defects of nesting behavior. These results indicate that the administration of EK100, antrodin C, and ibuprofen may have the potential to restore multiple brain injury in APP/PS1 mice.
Then, we studied the effects of four compounds on amyloid plaque deposition and Aβ perivascular deposition. EK100 and antrodin C reduced the number of plaques but did not reduce the levels of Aβ in the hippocampus and serum, indicating that the reduction in the number of plaques was not due to the inhibition of Aβ production. The use of SH-SY5Y-APP695 cells also confirmed the ineffective inhibition of Aβ accumulation by EK100 and antrodin C. Therefore, the effect of EK100 and antrodin C on reducing the number of plaques can be attributed to the removal of Aβ rather than the formation of Aβ.
Recent evidence suggests that impaired clearance may be the driving force behind sporadic AD [39]. Microglia may promote Aβ clearance through phagocytosis [40]. Although it is obvious that astrocytes and microglia accumulate around amyloid plaques in AD, it is still elusive whether they are mainly attracted by amyloid deposits or only by the plaque-related damaged neurite response [41]. Microglia activation is highly correlated with the accumulation of Aβ, because activated microglia are found to surround the plaque. Therefore, Aβ can be cleared by phagocytosis or proteolytic degradation [42]. Studies have shown that limiting the accumulation and phagocytosis of microglia will increase Aβ deposition, thus highlighting the functional impact of phagocytosis [43].
It is worth noting that EK100, antrodin C, and ibuprofen can effectively improve nesting behavior and inhibiting NCM activation, indicating that these two events are related in APP/PS1 mice. Microglia exhibit a variety of phenotypic states, from proinflammatory M1 phenotype to the alternative activation M2 phenotype, especially under chronic inflammatory conditions [44]. In vitro evidence suggests that the phagocytic ability of microglia is inhibited in AD [45]. The activity of M1-like reactive microglia induced by LPS in Aβ phagocytosis is significantly reduced, and this reduction can be rescued by IL-4 induced activation of M2-like microglia [45]. This phenomenon leads to the hypothesis that the accumulation of Aβ in AD may be due to changes in the phenotype of microglia. Therefore, the regulation of M2-like reactive microglia by microglia may have potential benefits in the treatment of AD.
Nrf2 is overexpressed in the cerebral cortex neurons of APP/PS1 mice, indicating that the overexpression of Nrf2 may be related to the antioxidant pathway of Nrf2 in neurons, but it has nothing to do with microglia. The overexpression of Nrf2 can be down-regulated by EK100, antrodin C, and ibuprofen, indicating that the overexpression of Nrf2 may be a feedback effect to protect neurons after the presence of Aβ. The treatment of EK100, antrodin C, and ibuprofen may overcome the toxicity mediated by Aβ. Therefore, the feedback effect of Nrf2 can be avoided.
The Nrf2/HO-1/NQO-1 pathway in microglia can regulate the inflammatory function of microglia and inhibit Aβ accumulation through phagocytosis [46]. The oxidation and anti-oxidation mechanisms are usually balanced by certain known elements, such as Nrf2 and HO-1. The overproduction of reactive oxygen species (ROS) and/or inhibition of antioxidant defense mechanisms may become harmful, which is called oxidative stress [47]. Previous studies reported that despite the presence of oxidative stress, the expression of nuclear Nrf2 in the brains of human AD patients is reduced [48]. However, other studies have shown that the expression of Nrf2 target genes in AD brain is increased. Previous studies have shown that Nrf2 transcripts are significantly reduced in 3-month-old APP/PS1 mice but not found in 6-month-old mice [49]. Fragoulis et al. demonstrated that in the AD mouse model, the administration of methysticin activates the Nrf2 pathway and reduces neuroinflammation, hippocampal oxidative damage, and memory loss [50]. Tanji et al. demonstrated that the expression level of HO-1 in the temporal cortex of AD patients is increased compared with the control group [51]. Other studies have reported increased NQO1 activity and immunoreactivity in the brains of AD patients associated with AD pathology [52]. A reasonable explanation for the differences in these reports is that Nrf2 levels may change during disease progression based on the degree of ROS production. It is now accepted that Nrf2 is up-regulated in the early stages of AD by Aβ-induced ROS but starts to decrease as the disease progresses [53]. Kanninen et al. used APP/PS1 mice to prove that Nrf2 expression decreases in the later stage [54]. Although the underlying mechanism has not been elucidated, the damage of the Nrf2 pathway may be related to the progression of the disease. Yammzaki et al. found that the expression of Nrf2 increased in APP/PS1 mice [55]. This may be mediated by the Aβ-mediated phosphorylation of P62, which interacts with Kelch-like ECH-associated protein 1 (Keap1). In addition, cell type-specific Nrf2 expression should be checked [56].
Aβ clearance can be mediated by microglia. In addition, Aβ can also be cleared through vascular access [57], the glymphatic system [58], and IPAD [59]. IPAD failure can lead to cerebral amyloid angiopathy (CAA), where Aβ mainly accumulates in capillaries and arterial smooth muscle cells BM [4]. Therefore, the perivascular deposition of Aβ was determined. The results showed that EK100 and antrodin C significantly increased the perivascular deposition of Aβ, indicating that EK100 and antrodin C reduced plaque deposition by promoting the clearance of Aβ through the perivascular pathway. However, partial failure of these clearance pathways can increase perivascular deposits.
It has been found that antrodin C has anti-inflammatory effects in RAW264.7 macrophages activated by LPS [20] and can effectively interfere with hyperglycemia-induced senescence and apoptosis by activating the HO-1/NQO-1-dependent cellular antioxidant defense system [21]. In our current study, the promotion of HO-1/NQO-1-dependent cellular antioxidant defense was detected in the treatment of antrodin C and ergosterol, but it was not detected during the treatment of EK-100 and ibuprofen. This may help antrodin C to reduce Aβ burden more than EK100.
EK100 also has anti-inflammatory effects [17] and has been studied to improve ischemic stroke [18]. In our current study, EK100 did not show cellular antioxidant defense that depends on HO-1/NQO-1. Indomethacin was used as a positive control to evaluate the analgesic activity of EK100 in vitro [60]. Our current study uses ibuprofen as an NSAID control and found that EK100 has a similar effect on NCA activation as ibuprofen, which means that the NSAID activity of ibuprofen and EK100 can increase the activation of NCA. In previous studies, it has been found that EK100 and its isomer ergosterol can inhibit LPS-mediated macrophage activation [17]. In our current study, we also found that both of these compounds can inhibit LPS-mediated BV2 microglia activation. However, EK100 (instead of ergosterol) inhibits the activation of non-clustered and clustered microglia, indicating that Aβ-mediated activation of microglia is specifically inhibited by EK100 but not affected by ergosterol.
It is speculated that both soluble Aβ and Aβ plaques can cause neuroinflammation and subsequent damage to the hippocampal nerves, which then leads to behavioral defects. EK100 can effectively remove plaque, reduce neuroinflammation, increase neurogenesis, and improve behavioral defects. According to this interference, antrodin C may also have the ability to promote neurogenesis.
In this study, we revealed four compounds with anti-inflammatory activity, two of which are structurally similar and exhibit different effects on AD-related pathological changes (including brain Aβ clearance). However, we cannot determine the precise target of these effects by using APP/PS1 mice as an animal model. In addition, the structureactivity relationship between EK100 and ergosterol and between antrodin C and ibuprofen has not been resolved. The ultimate goal of this study is to reveal the Aβ clearance promoting activity and the structure-activity relationship of these compounds. The biggest challenge in the future is that we need more technologies and research models, including cell and animal models, to achieve this goal.

Extraction, Isolation, Purification, and Structure Determination of Compounds in A. cinnamomea
Freeze-dried powder of A. cinnamomea of the submerged whole broth (Batch No. MZ-247) was provided by the Biotechnology Center of Grape King Inc., Chung-Li City, Taiwan, Republic of China. The purification procedure of EK100 from A. cinnamomea was described previously [61]. In brief, freeze-dried powder of A. cinnamomea of the submerged whole broth was extracted three times with methanol at room temperature. The methanol extract was evaporated in vacuum to give a brown residue, which was suspended in H 2 O and then partitioned with 1 L of ethyl acetate. The ethyl acetate fraction was chromatographed on silica gel using mixtures of hexane and ethyl acetate of increasing polarity as eluents and further purified with high-performance liquid chromatography (HPLC). Twelve components were identified. EK100 was eluted with 10% ethyl acetate in hexane. The structure of EK100 was elucidated by mass and nuclear magnetic resonance (NMR) spectral data.
The purification procedure of antrodin C from A. cinnamomea mycelia was described previously [17]. In brief, the dried and ground mycelium of A. cinnamomea was extracted with 95% ethanol. Then, the 95% ethanol extract was concentrated under reduced pressure. The residues are suspended in H 2 O and partitioned with n-hexane and then ethyl acetate. The ethyl acetate fraction was sequentially chromatographed on silica gel and Sephadex LH-20 columns to obtain antrodin C. The structure of antrodin C was elucidated by mass and NMR spectral data.

Cell Culture
BV2 cells, a cell line derived from primary mouse microglia cells, were maintained in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 µg/mL streptomycin. For treatment, the cell medium was replaced with DMEM containing 1% FBS.

Measurement of Nitric Oxide
For the assessment of the amount of nitric oxide production, the Griess reagent (0.05% N-(1-naphthyl)-ethylene-diamine dihydrochloride, 0.5% sulfanilamide, and 1.25% phosphoric acid) was employed. The accumulated nitrite, a stable breakdown product of nitric oxide, can be recorded. The optical density was detected at a wavelength of 540 nm using a microplate reader with NaNO 2 as standard.

Animal Management and Administration
The Institutional Animal Care and Use Committee (IACUC) at the National Research Institution of Chinese Medicine approved the animal protocol (IACUC No: 106-417-4). All experimental procedures involving animals and their care were carried out in accordance with The Guide for the Care and Use of Laboratory Animals published by the United States National Institutes of Health. APP/PS1 was purchased from Jackson Laboratory (No. 005864). The breeding gender ratio was a male with two females in one cage. Experiments were conducted using wild-type siblings and APP/PS1 transgenic C57BL/6J mice. The animals were housed under controlled room temperature (24 ± 1 • C) and humidity (55-65%) with a 12:12 h (07:00-19:00) light-dark cycle. All mice were provided with commercially available rodent normal chow diet and water ad libitum.
The dose of EK100 used for humans is typically 7380 mg per day. The mouse dose used was converted from a human-equivalent dose (HED) based on body surface area according to the US Food and Drug Administration formula. Assuming a human weight of 60 kg, the HED for 7380 (mg)/60 (kg) = 123 × 12.3 = 10 mg/kg; the conversion coefficient of 12.3 is used to account for differences in body surface area between mice and human, as previously described [63]. To investigate the effect of EK100 in CNS, 3 doses (30 mg kg −1 day −1 ) were applied. EK100, ergosterol, antrodin C, and ibuprofen were dissolved in vehicle (10% Kolliphor EL, 5% ethanol, 85% dextrose solution (5% in water), pH 7.2) to get a final concentration of 3 mg/mL. For studying the therapeutic effect, thirty APP/PS1 mice (5 months of age) were randomly assigned to five groups (n = 6, half male and half female) and were administrated by gavage with vehicle, EK100, antrodin C, ergosterol, and ibuprofen (30 mg·kg −1 ·day −1 ) for one month. Ergosterol was used as EK100's structural control and ibuprofen was used as an NSAID control. Nesting task was performed 30 days after drug administration. The animals were sacrificed on the 31st day after drug administration, and then, the Aβ-related pathological changes were inspected by immunohistochemistry and immunoblotting. For neurogenesis measurement, BrdU was injected intraperitoneally at 50 mg·kg −1 ·day −1 during the last 7 days.

Nesting Test
After oral gavage administration for 30 days, mice were assessed for a nesting test as described previously. In brief, two Nestlets (5 g) were placed into cage at 1 h before the dark cycle, and then, the nest score and the weight of unshredded Nestlets were determined after overnight. Nest construction was scored using a six-graded scale [64]. A score of 0 indicates undisturbed Nestlet; 1, Nestlet was disturbed, but nesting material has not been gathered to a nest site in the cage; 2, a flat nest; 3, a cup nest; 4, an incomplete dome and 5, a complete and enclosed dome.

Tissue Processing
Mice after being anesthetized were sacrificed by transcardial saline perfusion. The mouse brain was removed, and half of the brain was homogenized in homogenization buffer containing 20 mM Tris-HCl (pH 7.4), 320 mM sucrose, 2 mM ethylene diamine tetraacetic acid, 1 mM phenylmethylsulfonyl fluorid, 5 µg·mL −1 leupeptin, and 5 µg·mL −1 aprotinin. Another half brain was immersed in 4% formaldehyde overnight at 4 • C and cryoprotected. Then, brain tissue was sectioned into 30 µm thick sections. Three slides spanning approximately bregma-1.58 to -1.82 in each brain were used for staining and analysis.

Amylo-Glo Staining
Staining for fibrillary amyloid was performed using Amylo-Glo as described by the manufacturer (Biosensis Inc., Thebarton, South Australia).
For determining perivascular Aβ deposition, after merging the images of the GFAP and Aβ channels into one, the perivascular Aβ deposition was determined by calculating the positive pixels present in the GFAP vascular mask in the image.
Activated glia is a prominent feature of AD neuropathology, with both reactive astrocytes and activated microglia clustering around amyloid plaques [41]. Therefore, the number of clusters with plaque-associated astrocytes (PAA) and plaque-associated microglia (PAM) were determined by staining with GFAP-antibody and Iba-1-antibody, respectively. The reduced number of clusters with PAM may be related to the reduced number of plaques. To verify the relationship of cluster size with plaque size, the size alteration of the individual plaque-associated microglia cluster was analyzed by scatter plot.

Skelecton Analysis
The number of branches, junctions, and endpoints of microglia were quantified using FIJI ImageJ software [32]. Briefly, the images of brain sections were transformed to 8-bit format, and the Unsharp Mask and Despeckle were applied to increase contrast and remove noise; then, the images were skeletonized and analyzed using the plug-in AnalyzeSkeleton (2D/3D).

Quantification of AB10-Stained Plaques
The quantification of AB10-stained plaques was conducted. At least 3 coronal brain sections from each mouse were used for analysis. Each image was adjusted to the threshold for pixel detection (threshold setting for AB10-positive signal is 200). To eliminate background, particle large than 80 pixels (approximately 56 µm 2 ) less than 20 pixels (approximately 14 µm 2 ) was excluded.

Measurement of Aβ Levels
Two-step sequential extraction of the brain Aβ using 2% sodium dodecyl sulfate (SDS) and 70% formic acid (FA; Sigma) was processed as described previously [30]. Briefly, cortical homogenate was mixed with an equal volume of 4% SDS in homogenization buffer containing protease inhibitor. Then, the sample was sonicated and centrifuged at 100,000× g for 60 min at 4 • C. The supernatant was considered an SDS-soluble fraction. The SDS-insoluble pellet was further suspended in 70% FA and centrifuged at 100,000× g for 60 min at 4 • C. The supernatant was collected and neutralized with 1 M Tris, pH 11. SDSsoluble and SDS-insoluble fractions were stored at −80 • C until sandwich enzyme-linked immunosorbent assay (ELISA) analysis. Aβ level was measured by a sensitive sandwich ELISA assay using a kit (Invitrogen KHB3481 and KHB3442). The detailed experiments were performed according to the manufacturer's protocol.

Statistical Analysis
Results are expressed as mean ± standard error of the mean (S.E.M) and processed for statisyical analysis using GraphPad Prism 6 software (La Jolla, CA92037 USA). The parametric data were analyzed by unpaired two-tailed Student's t test or one way analysis of variance (ANOVA) with post hoc multiple comparisons with a Bonferroni test.

Conclusions
Our in vitro experiments have shown that EK100, antrodin C, and ergosterol have anti-inflammatory activity by detecting the production of nitric oxide. Ibuprofen is an NSAID that can specifically block the COX-2 enzyme. On the other hand, our in vivo experiments show that EK100 and antrodin C reduce the formation of amyloid plaques in the cortex and hippocampus, which may be due to the glial phagocytosis and perivascular granule pathways in APP/PS1 mice to promote Aβ clearance. We also showed that EK100 and antrodin C ameliorated behavioral deficits in APP/PS1 mice. EK100 also promotes hippocampal neurogenesis. These findings raise the possibility that EK100 and antrodin C may have the potential to treat AD.