Ginsenoside Rg1 alleviates Aβ deposition and neuronal damage by inhibiting NLRP1 inammasome activation in APP/PS1 mice

Background: Oxidative stress and neuroinammation play important roles in the whole pathogenesis of Alzheimer's disease (AD). NADPH oxidase 2 (NOX2) is an important enzyme that is responsible for ROS generation in many neurodegenerative diseases. The nucleotide-binding oligomerization domain (NOD)-like receptor protein 1 (NLRP1) inammasome is responsible for the formation of pro-inammatory molecules in neurons. However, it is still unclear whether inhibition of NOX2 and NLRP1 inammasome decreases amyloid-beta (Aβ) generation and deposition in APP/PS1 mice. Ginsenoside Rg1 (Rg1) is an active component in ginseng, and maybe a potential agent for neurodegenerative diseases. In this study, we investigated the effects and mechanisms of Rg1 treatment on cognitive performance, neuronal damage, Aβ deposition and NOX2-NLRP1 inammasome activation in APP/PS1 mice. Methods: WT and APP/PS1 mice were used in the experiment from 6 months (M) old to 9M old, and 6M APP/PS1 mice were used as pre-treatment controls. The open eld test (OFT) and Morris water maze (MWM) were used to detect behavioral change and cognitive function. The H&E and Nissl staining were used to assess neuronal damage. We further examined PSD95 expression, Aβ generation and deposition, Tau and p-Tau expression, NOX2-mediated ROS generation and NLRP1 inammasome activation by using immunouorescence, western blot analysis, and real time q-PCR. Results: Rg1 treatment for 12 weeks alleviated learning and memory impairments and neuronal damage, decreased p-Tau level, APP expression and Aβ deposition in APP/PS1 mice. Meanwhile, Rg1 treatment signicantly decreased the levels of ROS and IL-1β, and reduced the expressions of NOX2 and NLRP1 inammasome in the brain cortex and hippocampus in APP/PS1 mice. Furthermore, apocynin (a NOX inhibitor) and NLRP1-siRNA treatment also alleviated cognitive into Aβ. In the present study, the results showed that Rg1 treatment had no signicant effect on NCSTN expression, but could signicantly decrease the expression levels of APP, BACE and CTF-β, which were signicantly increased in APP/PS1-9M mice. Meanwhile, we also detected the expression of Aβ 1-42 , a pernicious peptide, and the levels of amyloid plaque in the cortex and hippocampus. The results showed that Rg1 treatment signicantly decreased the expression of Aβ 1-42 and the levels of amyloid plaque in APP/PS1 mice. Additionally, Rg1 treatment signicantly decreased the levels of p-Tau and senescence-associated β-Gal that were also signicantly increased in the APP/PS-9M mice. These results suggest that Rg1 may not only decrease APP expression but also inhibit the β-secretase pathway in APP processing to hinder pernicious Aβ formation.

Alzheimer's disease (AD) is a neurodegenerative disease characterized by progressive memory loss and cognitive decline. Clinically, the main pathological features of AD are the extracellular accumulation of senile plaques comprised of amyloid-beta (Aβ) peptides, intracellular neuro brillary tangle composed of highly phosphorylated Tau (p-Tau) and neurodegenerative changes accompanied by synaptic dysfunction and neuronal damage [1]. Currently, the pathogenesis of AD is still not completely understood. Increasing evidence suggests that neuroin ammation plays important role in the pathogenesis and the behavioral disturbances associated with AD [2][3][4]. Although a large number of epidemiological researches have proven that the traditional non-steroidal anti-in ammatory drugs (NSAIDs) could inhibit neuroin ammation and alleviate the severity of cognitive impairment, gastrointestinal complications limited the long-term use of these drugs [5]. At present, there are still no effective drugs for preventing the generation and progression of AD. Thus, the development of effective drugs with limited or no side effects is necessary for AD treatment.
In recent years the natural medicines with fewer side effects and safer performance have been widely used to treat many diseases including AD [6,7]. Ginseng has been safely used in China for more than 2000 years as it improves health conditions and delays senescence [8]. Ginsenoside Rg1 (Rg1), one of the main active components of ginseng, is a tetracyclic triterpenoid saponin [9]. It has been reported to be a potential neuroprotective agent in several animal models of neurological diseases and cognitive impairments [10,11]. Our previous study demonstrated that Rg1 was able to alleviate H 2 O 2 -induced neuronal damage by inhibiting NADPH oxidase 2 (NOX2) and nucleotide-binding oligomerization domain (NOD)-like receptor protein 1 (NLRP1) in ammasome activation in hippocampal neurons in vitro [12]. These data suggested that Rg1 treatment might delay the occurrence and development of AD through its antioxidant and anti-in ammatory effects. Therefore, it is important to study the protective effects and precise mechanisms of Rg1 on APP/PS1 AD mice.
It has been reported that oxidative stress plays an important role in the pathogenesis of many neurodegenerative diseases. Excessive reactive oxygen species (ROS) generation can not only induce oxidative stress damage but also activate an in ammatory response, which has been considered to contribute to the pathogenesis and progression of AD [13]. NADPH oxidases (NOXs) are the main enzymes that contribute to excessive ROS generation in many tissues and also involved in the development of the AD pathological process [14]. NOX2 is found to be constitutively expressed in the brain, especially in neurons [15]. A recent study suggested that NOX2 contributed to aging-related neuronal oxidative stress damage and loss of brain function [16]. Our previous research indicated that the expressions of NOX2 and its subunits of Rac1, p47phox and p22phox were signi cantly increased in long-term cultured hippocampal neurons [17]. Meanwhile, treatment with Rg1 signi cantly decreased the expressions of NOX2 and p47phox in H 2 O 2 -induced hippocampal neurons [12]. Therefore, we speculated that Rg1 treatment might inhibit NOX2-mediated neuronal oxidative stress and neuroin ammation in APP/PS1 AD mice.
Increasing evidence suggests that neuroin ammation is an underlying cause of many human diseases and plays a catalyst-like role in the development of AD [18,19]. The in ammatory response is not only a reaction to the disease process but also a contributor to the neuronal damage. The in ammasomes are multiprotein complexes in the cytoplasm that are responsible for the formation of pro-in ammatory molecules. The NLRP1, a critical component of the in ammasome, appears to be expressed ubiquitously, and a high level of NLRP1 is found in the brain, particularly in neurons [20,21]. A recent study reported that the NLRP1 immunopositive neurons in AD brains were increased by 25 to 30 folds compared with non-AD brains [21]. Similarly, inhibiting the expression of NLRP1 could reduce the Aβ deposition in the brain of AD model mice [22]. Our previous study showed that the NOX2 and NLRP1 in ammasome were signi cantly activated in prolong primary cultured hippocampal neurons [17]. However, it remains unclear whether Rg1 can improve cognitive function and ameliorate neuronal damage and Aβ deposition through inhibiting the NOX2-NLRP1 in ammasome activation in APP/PS1 AD mice.
The APP/PS1 model mice of AD exhibit remarkable elevation of Aβ production associated with certain behavioral abnormalities. It has been reported that the expression of NLRP1 in ammasome increased in the brain of APP/PS1 mice with increasing of age [22]. In the present study, we hypothesized that Rg1 treatment could alleviate Aβ generation and deposition by inhibiting NOX2 and NLRP1 in ammasome activation in the APP/PS1 AD mice. Apocynin, a NOX inhibitor, can inhibit the activity of NADPH oxidase by interfering with the intracellular translocation of its two cytoplasmic components of p47phox and p67phox [23]. It has been reported that apocynin has bene cial antioxidant effects on neurological diseases [24,25]. In this study, to con rm our hypothesis, we investigated the protective effects and mechanism of Rg1 and apocynin on the locomotor activity, cognitive ability, neuronal damage, Aβ deposition and NOX2-NLRP1 in ammasome activation in the APP/PS1 AD mice. Meanwhile, we further studied the effect of NLRP1 knockdown on cognitive impairment, neuronal damage and Aβ deposition in APP/PS1 AD mice.

Animals and treatment
The study was performed on APP/PS1 female mice and wild-type (WT) littermates. The APP/PS1 transgenic mice were purchased from the Model Animal Research Center of Nanjing University. Male APP/PS1 mice were used in the experiment, and age-and gender-matched WT littermates were used as controls. Mice were housed in standard laboratory conditions with free access to standard food and water. The mice were allowed to adapt to the laboratory conditions before testing. The experimental protocols were approved by the animal ethics committee of Anhui Medical University.
After the behavior test, the mice were decapitated and the brains were quickly removed for other experiments.

Open eld test (OFT)
The OFT equipment (Shanghai Biotechnology Co., Ltd.) includes a video-tracked cage (60 × 60 × 50 cm) and ANY-maze behavior tracking software (Stoelting Co., Wood Dale, IL, USA). The cage was divided into nine squares by two horizontal lines and two vertical lines as described in a previous study [28]. For OFT, each mouse was placed in the cage for 2 minutes to adapt to the environment. Then the ANY-maze behavioral tracking software was used to record the motor activity for 3 min. The total moving distance (m), the mean movement speed (m/s), the number of lines crossed, and the number of stand up were calculated by the software to indicate the motor activity behavior. In the OFT, each group except the APP/PS1-6M group was evaluated motor activity changes at 6 months old and every 4 weeks thereafter.

Buried Food Test (BFT)
In recent years, it has been reported that olfactory dysfunction occurs in AD patients, even in an early stage of AD [29]. Therefore we further detected the change of olfactory dysfunction after the OFT experiment using the previously described method [30]. Brie y, at 24 h prior to testing, the mice were subjected to a food-restricted diet. All mice were habituated to the testing room for 1 h prior to testing, then the mouse was allowed to adapt to the cage for 5 min. A small piece of food was randomly placed 1 cm beneath the bedding in a corner of a clean testing cage with 3 cm of woodchip bedding. The mouse was then placed in the testing cage at a constant distance from the hidden food. The ANY-maze behavioral tracking software was used to record the latency (s) of the mouse to nd the food. If the mouse failed to nd the buried food within 5 min, the latency score was recorded as 300 s. Nine hours after the above experiment, the surface pellet test was performed using the same scheme but the pellet was placed on the surface to exclude possible motor disorders.

Morris water maze (MWM) test
MWM is an important method to detect learning and memory impairments [31]. The MWM test includes four consecutive daily training trials and the fth day of exploration trials as described in our previous study [32]. The mean escape latency (MEL, s) of the four trials each day was recorded to indicate the learning performance. On the fth day, the platform was removed from the pool, and each mouse performed a 60s swimming probe trial. The latency of rst entry to the platform (LFP, s), the swimming time in the quadrant of the platform (STP, s), and the number of crossing the platform (NCP) were recorded to indicate the memory results.

Histological examination
After the MWM test, the mice were sacri ced and were perfused transcardially with 4% paraformaldehyde to x the brains. Then para n-embedded brain tissues were sliced into 5 μm sections using a microtome (Leica, Nussloch, Germany). The sections were stained with hematoxylin and eosin (H&E) to examine the pathological changes in cortex and hippocampus by intelligent tissue slice imaging analysis system platform (Pannoremic MIDI, 3DHISTECH, Hungary ). Besides, Nissl staining is often used to identify neuronal damage [33]. We further performed the Nissl staining to study the neuronal pathological alterations. The sections were treated with Nissl staining solution (Beyotime Biotechnology, China) for 10 min and were differentiated with 95% alcohol. Then the sections were observed and photographed by a light microscope (Olympus IX71, Japan). The density of the Nissl bodies was analyzed double-blindly from three random elds (400×) in the cortex and hippocampal CA1 and CA3 regions in each section by using the Image-Pro Plus 6.0 automatic analysis system to assess the amount of Nissl bodies.

Determination of ROS production
For analysis of brain ROS production, the dihydropyridine (DHE) uorescence staining was performed to examine the ROS levels in the brain [34]. Brie y, DHE (Beyotime Biotechnology, Shanghai, China) was injected from the tail vein (100 μM, 0.1 ml/10g body weight) for 30 minutes. Then the brains were removed and embedded in OCT (Sakura, Torrance, CA) at −20 °C. The brain was cut into 10 μm sections by using a freezing microtome (Leica CM3050, Germany). Then the sections were washed and stained with DAPI for 5 minutes and were photographed by using a uorescence microscope (Olympus IX71). The exposure time was 900 ms for all sections. The mean uorescence density was quanti ed doubleblindly from three random elds (400×) in the cortex, hippocampal CA1 and CA3 in each section by using Image-Pro Plus 6.0 automatic analysis system to assess the ROS production.
Determination of senescence-associated β-Galactosidase (β-Gal) activity The β-galactosidase (β-Gal) is frequently used as a marker of cellular senescence. The frozen brain sections (5 μm) were stained with a β-Gal kit (Beyotime Biotechnology, Shanghai, China) to detect the β-Gal activity. Brie y, the sections were incubated with the β-Gal staining solution overnight at 37 °C. Blue products were generated under the catalysis of β-Gal with X-Gal as the substrate. The results were viewed by using a microscope (Olympus IX72, Japan). The mean density of the blue cells was quanti ed double-blindly from three random elds (400×) in the cortex, hippocampal CA1 and CA3 in each section using Image-Pro Plus 6.0 automatic analysis system to assess the β-Gal activity.

Immuno uorescence
The para n sections were used to perform immuno uorescence. Brie y, the sections were depara nized and hydrated. Then the sections were permeabilized with Triton X-100 (0.25%) for 30 min, followed by blocking with BSA (1%) for 60 min. Then the sections were incubated with primary antibodies overnight at 4 °C. The antibodies were rabbit polyclonal postsynaptic density 95 (PSD95) (1:100, Servicebio Technology, GB11277), rabbit polyclonal Aβ 1-42 (1:150, Bioss Technology, bs-0076R). The next day, the sections were incubated with secondary antibodies against mouse or rabbit which were conjugated to Rhodamine (1:200, ZSGB-BIO). Then the sections were mounted by using an anti-fade medium and examined with the uorescence microscope (Olympus IX71, Japan). The uorescence density was analyzed double-blindly from three random regions (400×) of the cortex and hippocampus regions by using Image-Pro Plus 6.0 automatic analysis system to indicate the changes of PSD95 and Aβ  expression.
Thio avin-S Staining Para n sections (n = 5) were depara nized with xylene and were hydrated by graded alcohol. Then the sections were immersed in 1% Thio avin-S (HY-D0972, MedChemExpress USA) for 5 minutes, hydrated in 70% ethanol and washed twice with PBS (5 minutes each time). After staining with Hochest 33258 for 10 minutes, the slides were sealed with an anti-fade medium. The intelligent tissue slice imaging and analysis system platform (Pannoremic MIDI, 3DHISTECH, Hungary) was used to scan the images and analyze the data. The QuantCenter of Pattern of the platform is used to analyze the amyloid deposition areas and the hippocampus and cortex areas in each section. The percentage of Thio avin-S positive areas was counted to evaluate changes in amyloid deposits.

Immunoblot analysis
The hippocampus and cortex tissues were homogenized in RIPA lysis (Beyotime Biotechnology, Shanghai, China) with protease and phosphatase inhibitors. Total protein was extracted using a cryogenic tissue grinder (Shanghai Jingxin Industrial Development Co., Ltd) at 4 ° C, 60 Hz for the 50s.
The BCA Protein Assay Kit (Beyotime Biotechnology, Shanghai, China) was used to determine protein concentration. The samples (n = 4 per group) were separated by SDS-PAGE (8%-12%) and transferred onto PVDF membranes (Millipore, Bedford, MA, USA). Afterward, membranes were blocked in 5% non-fat milk in Tween20 TBS (TBST) for 1 h at room temperature, followed with the primary antibodies (Supplementary Table S1) overnight at 4°C. The next day, membranes were washed and incubated with a horseradish peroxidase-conjugated secondary antibody (ZSGB-BIO, ZF-2301, 1:10,000) for 1 h at room temperature. After washing three times with TBST, the proteins were visualized by an enhanced chemiluminescent reagent (Amersham Biosciences, UK). The bands were obtained by using a Chemi Q4800 mini imaging system (Shanghai Bioshine Technology, Shanghai, China). The density of the protein band was measured using Image J 6.0. The density of the target band was normalized to β-actin. The relative density of target protein over WT control was used to indicate the changes in target proteins.

Quantitative real-time PCR (q-PCR) analysis
For the q-PCR analysis, total RNA was extracted from the hippocampus and cortex tissues (n = 4) with TRIzol reagent (Invitrogen Co., USA) according to the manufacturer's instructions as described previously [35]. The rst-strand cDNA was synthesized from total RNA with PrimeScript™ Reverse Transcriptase (Takara Bio, RR037A) according to the manufacturer's protocol. Quantitative real-time PCR analyses for mRNAs of NLRP1, ASC, Caspase-1, IL-1β, NOX2, p22phox, p47phox, APP, BACE, NCSTN, and β-actin were performed with SYBR®Premix Ex Taq™II RTPCR kits (Takara Bio, RR820A). The mRNA level of β-actin was used as an internal control. Primers are listed in Supplementary Table S2. PCR was performed at 95°C for 30 s, followed by 40 cycles of ampli cation at 95 °C for 5 s, 60 °C for 30 s with Real-time PCR System (BIO-RAD, CFX96). The uorescent signals were collected during the extension phase, CT values of the sample were calculated, and transcript levels were analyzed by 2 −ΔΔCT method. The q-PCR was repeated three times.

Statistical analysis
All data are expressed as mean ± SD. Statistical analysis was performed using GraphPad Prism 6.0 software. Repeated measures analysis of variance (ANOVA) was used to analyze the results of OFT from 6-9 months and MWM training from D1-4, and other data were analyzed using one-way ANOVA. Student's t-test was used to compare the differences between groups. Statistical signi cance was de ned as P < 0.05.

Rg1 treatment ameliorates abnormal behaviors in APP/PS1 mice
In this study, the open eld test (OFT) was used to observe the effects of Rg1 treatment on motor behavior in APP/PS1 mice. The results showed that there were no signi cant differences between the WT-6M and APP/PS1-6M groups before Rg1 and apocynin treatment. While compared with 6M group, there were a decreasing trend in the moving distance (m), the mean moving speed (m/s), the number of lines crossing and the stand up with the age from 6M to 9M in WT control group and there were signi cant decrease in 8M and 9M ( Fig. 1A-D, P < 0.05 or P < 0.01). In the APP/PS1-9M model group, these parameters also showed a decreasing trend, but the difference has no signi cant ( Fig. 1A-D, P > 0.05). And in the apocynin and Rg1 (5 and 10 mg/kg) groups, similar to WT control group, these parameters also had a signi cant decrease, especially in 8M and 9M old ( Fig. 1A-D, P < 0.05 or P < 0.01). Meanwhile, the results in 9M showed that, compared with WT-9M control, the moving distance (m) and mean moving speed (m/s) were signi cantly increased in APP/PS1-9M group, and the Rg1 (10 mg/kg) could signi cantly decrease the moving distance and mean moving speed compared with APP/PS1-9M group ( Fig. 1A and B, P < 0.01). The number of lines crossing and stand up also decreased in the Rg1-treated groups but there were no statistical differences ( Fig. 1C and D). The data suggested that, with the increase of age, the APP/PS1 mice showed signi cant motor hyperactivity, Rg1 treatment signi cantly decreased the motor hyperactivity in the APP/PS1 mice.
We further used the BFT to detect the effect of Rg1 treatment on the alteration of olfactory function in APP/PS1 mice. The buried pellet results showed that there were no differences in the latency to nd the food between the WT-9M and APP/PS1-6M group mice, while the latency was signi cantly increased in the APP/PS1-9M group mice compared with WT-9M group mice. Rg1 and apocynin treatment could signi cantly decrease the latency to nd the food (Fig. 1E, P <0.05 or P < 0.01). Meanwhile, the surface pellet test showed similar results to the buried pellet test (Fig. 1F, P < 0.01), but the latency to nd the food was signi cantly shortened in all groups mice, suggesting there was no motor dysfunction in the experimental animals. These results suggested that Rg1 treatment could signi cantly improve the olfactory dysfunction in APP/PS1 mice.
We also use the MWM test to detect the effect of Rg1 treatment on learning and memory impairments in the APP/PS1 mice. The results of training trials showed that the escape latency has no signi cant difference in the rst day (D1) among these experimental groups. However, compared with the D1, the escape latency in WT-9M group was signi cantly reduced in the D3 and D4 (Fig. 1G, P < 0.01), the APP/PS1-6M group also had a decreasing trend and signi cantly reduced in D4 (Fig. 1G, P < 0.05), while the escape latency had no signi cant decrease in APP/PS1-9M group (Fig. 1G, P > 0.05). Similarly to WT-9M group, the escape latency in apocynin and Rg1 treatment groups also had a decreasing trend and signi cantly reduced in D4 (Fig. 1G, P < 0.05 or P < 0.01). Additionally, compared with the WT-9M group, the escape latency of APP/PS1-9M group was signi cantly increased in D2, D3 and D4 (Fig. 1G, P < 0.05 or P < 0.01). While compared with APP/PS1-9M group, treatment with apocynin and Rg1 (10 mg/kg) could signi cantly reduce the escape latency in D4 (Fig. 1G, P < 0.05 or P < 0.01). In the fth day of probe test, the results showed that, compared with WT-9M or APP/PS1-6M group, the latency of rst entry to the platform (LFP, s) was signi cantly increased ( Fig. 1H, P < 0.01), the number of crossing the platform (NCP) (Fig. 1I, P < 0.01) and the swimming time in the quadrant of platform (STP, s) ( Fig. 1J, P <0.05 or P <0.01) were signi cantly decreased in APP/PS1-9M mice. While compared with APP/PS1-9M group, treatment with apocynin and Rg1 (5 and 10 mg/kg) could signi cantly reduce the LFP, and increase the NCP and STP (Fig. 1H-J, P < 0.05 or P < 0.01). The results indicated that Rg1 treatment could signi cantly improve the learning and memory impairments in APP/PS1 mice.

Rg1 treatment alleviates neuronal damage in APP/PS1 mice
We observed the pathological changes using H&E and Nissl staining to evaluate the effect of Rg1 treatment on neuronal damage in APP/PS1 mice. The H&E results showed that there were no abnormal changes in the cortex and hippocampus in the WT-9M group mice. Compared with WT-9M group, there were a few neuronal pathological changes in the APP/PS1-6M group mice such as unclear nucleoli, karyopyknosis and amyloid plaque (as shown in red square), indicating that there were slight neuronal damages in cortex and hippocampus in the APP/PS1-6M mice. While in APP/PS1-9M mice, there were obvious neuronal damage and amyloid plaque in the cortex and hippocampus. And compared with APP/PS1-9M mice, apocynin and Rg1 treatment could signi cantly reduce the neuronal damages and amyloid plaque in the cortex and hippocampus ( Fig. 2A). The data suggested that Rg1 treatment could signi cantly alleviate the neuronal damage in the APP/PPS1 mice.
Nissl staining results showed that there were abundant Nissl bodies in the cortex and hippocampal neurons in the WT-9M and APP/PS1-6M group mice. Compared with WT-9M group, the Nissl bodies were signi cantly decreased in hippocampus CA3 and cortex in APP/PS1-9M group mice ( Fig. 2B and C, P < 0.05 or P < 0.01). While compared with APP/PS1-9M group, treatment with apocynin and Rg1 could signi cantly increase the Nissl bodies in the hippocampus CA3 and cortex in the APP/PS1 mice, especially in Rg1 10 mg/kg group ( Fig. 2B and C, P < 0.05 or P < 0.01). The results indicated that Rg1 treatment had a protective effect on neuronal damage in APP/PS1 mice. β-gal is also an important biomarker for neuronal aging. β-gal activity is often measured to evaluate the age-related neuronal damage [36]. In this study, we further examined the expression of β-gal by immunoblot. The results showed that there were few expression of β-gal in WT-9M and APP/PS1-6M group mice, while compared with WT-9M or APP/PS1-6M group, the expression of β-gal was signi cantly increased in APP/PS1-9M group ( Fig. 2D and E, P < 0.01). And compared with APP/PS1-9M group, treatment with apocynin and Rg1 could signi cantly reduce the expression of β-gal in APP/PS1 mice ( Fig.  2D and E, P < 0.01). Additionally, we also detected the β-gal activity by β-gal staining. The results are similar to the results of immunoblot (Fig. S1A-D). These results suggested that Rg1 treatment could alleviate aging-related neuronal damage in APP/PS1 mice.
Behavioral dysfunction usually accompanied by changes in synaptic structure and function [37]. To evaluate the effect of Rg1 treatment on changes of synaptic function, we further detected the expression of PSD95 by immuno uorescence and immunoblot. The immuno uorescence results showed that there was abundant PSD95 expression in the WT-9M group. Compared with WT-9M group, the expression of PSD95 decreased in APP/PS1-6M group, but there was no signi cant differences. While in APP/PS1-9M group, the expression of PSD95 was signi cantly reduced ( Fig. 3A-D, P < 0.01). Compared with APP/PS1-9M, treatment with apocynin and Rg1 could signi cantly increase the expression of PSD95 in APP/PS1 mice ( Fig. 3A-D, P < 0.01). The immunoblot results were similar to immuno uorescence in PSD95 expression ( Fig. 3E and F, P < 0.01). The results indicated that Rg1 treatment might improve the synaptic dysfunction in APP/PS1 mice.

Rg1 treatment reduces Aβ deposition and Tau hyperphosphorylation in APP/PS1 mice
The amyloid-β deposition is closely related to the generation and progression of pathological characteristics in APP/PS1 mice [38]. In this study, we rst detected the expressions of APP and its metabolizing enzymes by immunoblot. The results showed that, compared to WT-9M group, the expression levels of APP, CTF-β and BACE had no signi cant difference in APP/PS1-6M group mice, while in APP/PS1-9M group, they were signi cantly increased (Fig. 4A-D, P < 0.01). Compared with APP/PS1-9M group, treatment with apocynin and Rg1 could signi cantly decrease the expression levels of APP, CTF-β and BACE in APP/PS1 mice (Fig. 4A-D, P < 0.05 or P < 0.01). The results also showed that the expression of NCSTN had no signi cant changes in all groups mice (Fig. 4A and E). We also further detected the mRNA levels of APP, BACE and NCSTN. The results showed that the mRNA levels of APP, BACE and NCSTN were similar to their protein expression (Fig. 4F-H, P < 0.05 or P < 0.01). These data suggested that Rg1 treatment could signi cantly decrease the expressions of APP and BACE, and do not affect NCSTN expression.
To further con rm the protective effect of Rg1 on APP/PS1 mice, we detected the Aβ deposition by Thio avin-S staining and expressions of Aβ 1-42 by immuno uorescence and immunoblot. The Thio avin-S staining results showed that there were few Aβ deposition in the cortex and hippocampus in WT-9M group mice. And compared with the WT-9M group, the Aβ deposition had a slight increase in the APP/PS1-6M group mice, while in APP/PS1-9M group, the Aβ deposition was signi cantly increased in the cortex and hippocampus ( Fig. 4I and J, P < 0.01). Compared with APP/PS1-9M group, treatment with apocynin and Rg1 could signi cantly decrease the Aβ deposition both in the cortex and hippocampus ( Fig. 4I and J, P < 0.05 or P < 0.01). Additionally, to con rm the effect of Rg1 treatment on Aβ deposition in APP/PS1 mice, we further detect the expression of Aβ 1-42 using immuno uorescence. The results were similar to the Thio avin-S staining of Aβ deposition ( Fig. S2A and B, P < 0.01). Meanwhile, the immunoblot results also showed a similar result in the expression of Aβ 1-42 ( Fig. 4K and L, P < 0.01).
These results suggested that Rg1 could signi cantly inhibit Aβ deposition and reduce the formation of amyloid plaque.
In addition, we also observed the effect of Rg1 treatment on the expression levels of Tau and p-Tau. The results showed that there were no signi cant changes in the expression of Tau among all groups. While compared with WT-9M group, the expression level of p-Tau was signi cantly increased in APP/PS1-9M group mice ( Fig. 4M and N, P < 0.01). Compared with APP/PS1-9M group, treatment with apocynin and Rg1 could signi cantly reduce the expression level of p-Tau ( Fig. 4M and N, P < 0.05). The results suggested that Rg1 treatment might also alleviate the neuro lament tangle induced by hyperphosphorylation of Tau.

Rg1 treatment decreases ROS generation and NOX2 expression in APP/PS1 mice
The accumulation of ROS plays a vital role in promoting the pathogenesis of AD [2,3]. We further detected the ROS production in the cortex and hippocampus in the APP/PS1 mice. The results showed that there was slight ROS production in the cortex and hippocampal CA1 and CA3 areas in WT-9M and APP/PS1-6M group mice, while the ROS production was signi cantly increased in the cortex and hippocampus in APP/PS1-9M group compared with the WT-9M group (Fig. 5A-D, P < 0.01). Compared with APP/PS1-9M, apocynin and Rg1 treatment could signi cantly reduce the ROS production in the cortex and hippocampal CA1, CA3 areas ( Fig. 5A-D, P < 0.01).
We further detected the expressions of NADPH oxidase 2-related protein and mRNA of NOX2, p22phox and p47phox, which were closely involved in ROS generation in neurons. The immunoblot results showed that, compared with WT-9M group, the expression levels of NOX2, p22phox and p47phox had no signi cant increase in the APP/PS1-6M group mice, while they were signi cantly increased in the APP/PS1-9M group mice (Fig. 5E-H, P < 0.01). Compared with the APP/PS1-9M group, apocynin and Rg1 treatment could signi cantly decrease the expression levels of NOX2, p22phox and p47phox (Fig. 5E-H, P < 0.01). Meanwhile, the q-PCR results showed that, similar to their proteins, the levels of NOX2, p22phox and p47phox mRNA were signi cantly increased in APP/PS1-9M group mice (Fig. 5I-K, P < 0.01), and were signi cantly decreased after treatment with apocynin and Rg1 in APP/PS1 mice (Fig. 5I-K, P < 0.01). These data suggested that Rg1 treatment might reduce ROS generation by inhibiting NOX2 expression in the APP/PS1 mice.
Rg1 treatment decreases the expression of NLRP1 in ammasome in APP/PS1 mice It has been reported that NOX2 plays an important role in the activation of NLRP1 in ammasome [26].
Therefore, we further detected the expression of NLRP1 in ammasome-related proteins and mRNA of NLRP1, ASC, caspase-1 and IL-1β, and detected the expression levels of NF-κB and p-NF-κB by immunoblot. The results showed that, compared with WT-9M group, there was no signi cant elevation in the protein and mRNA of NLRP1, ASC, caspase-1 and IL-1β in APP/PS1-6M group mice, while in APP/PS1-9M group mice, the expression levels of protein and mRNA of NLRP1, ASC, caspase-1 and IL-1β were signi cantly increased (Fig. 6A-I, P < 0.05 or P < 0.01). Compared with APP/PS1-9M group, treatment with apocynin and Rg1 could signi cantly reduce the expression levels of protein and mRNA of NLRP1, ASC, caspase-1 and IL-1β ( Fig. 6A-I, P < 0.05 or P < 0.01). The NF-κB and p-NF-κB results showed that there was no signi cant difference between the WT-9M and APP/PS1-6M groups. While compared with the WT-9M or APP/PS1-6M groups, the expression levels of NF-κB and p-NF-κB were signi cantly increased in the APP/PS1-9M group mice (Fig. 6J-L, P < 0.01). And compared with APP/PS1-9M group, treatment with apocynin and Rg1 signi cantly reduced the expression levels of NF-κB and p-NF-κB in APP/PS1 mice ( Fig. 6J-L, P < 0.05 or P < 0.01). These results suggested that Rg1 treatment could alleviate the neuroin ammation by inhibiting NLRP1 in ammasome activation in the APP/PS1 mice. NLRP1-siRNA treatment ameliorates abnormal behaviors in APP/PS1 mice To con rm the effect of NLRP1 in ammasome activation on learning and memory impairments and Aβ deposition in APP/PS1 mice, we further observed the effect of LV-NLRP1-siRNA treatment for 12 weeks on cognitive dysfunction, neurodegeneration and Aβ deposition. We rstly examined the effect of NLRP1-siRNA treatment on the motor behavior in APP/PS1 mice by using the OFT. The results showed that there were no signi cant differences among the NS, LV-scramble and LV-NLRP1-siRNA groups at 6M before treatment. While compared with 6M group, there was a decreasing trend in the motor behavior from 6M to 9M in all groups mice, especially in the LV-NLRP1-siRNA group, the moving distance (m), the mean moving speed (m/s), the number of lines crossing and the stand up were signi cantly decreased at 9M ( Fig. 7A-D, P < 0.05 or P < 0.01). Additionally, the results of 9M showed that, compared with NS control, LV-NLRP1-siRNA treatment could decrease the number of lines crossing and stand up behavior ( Fig. 7C and D, P < 0.05). These data suggested that NLRP1-siRNA treatment could decrease the hyperactivity behavior which increased in the APP/PS1 mice.
We further used the BFT to detect the effect of NLRP1-siRNA treatment on the alteration of olfactory function in APP/PS1 mice. The results showed that NLRP1-siRNA treatment had no signi cant effect on the latency to nd the food in both the buried pellet and surface pellet test ( Fig. 7E and F, P > 0.05). The results suggested that NLRP1-siRNA treatment did not affect the olfactory dysfunction in the APP/PS1 mice.
We also use the MWM test to detect the effect of NLRP1-siRNA treatment on learning and memory impairments in the APP/PS1 mice. The results of training trials showed that the escape latency has no signi cant difference in the rst day (D1) and D2 among the NS, LV-scramble and LV-NLRP1-siRNA treatment groups. However, compared with NS control group, the escape latency was signi cantly reduced in the D3 and D4 in the NLRP1-siRNA treatment group (Fig. 7G, P < 0.05). On the fth day of the probe test, the results showed that, compared with NS control group, NLRP1-siRNA treatment could signi cantly reduce the LFP, and increase the NCP and STP in APP/PS1 mice ( Fig. 7H-J, P < 0.05). The results indicated that NLRP1-siRNA treatment signi cantly alleviates the learning and memory impairments in APP/PS1 mice.

NLRP1-siRNA treatment inhibits NLRP1 in ammasome activation in APP/PS1 mice
We rstly con rmed the transfected effect of lateral ventricles injection of lentivirus with GFP. The results showed that treatment with lentivirus for 12 weeks showed obvious expression of GFP in hippocampus CA1, CA3 and cortex, indicating successful lentivirus infection in cortex and hippocampus in APP/PS1 mice (Fig. S3). Meanwhile, we further detected the effect of NLRP1-siRNA treatment on the expression levels of NLRP1-related protein and mRNA. The results showed that there were no signi cant differences between the NS and LV-scramble groups. While compared with NS or LV-scramble treatment group, NLRP1-siRNA treatment for 12 weeks could signi cantly decrease the protein and mRNA levels of NLRP1, ASC, caspase-1 and IL-1β which increased in APP/PS1 mice ( Fig. 8A-I, P < 0.01). We also detected the expression levels of NF-κB and p-NF-κB. The results showed that, compared with NS or LV-scramble treatment group, NLRP1-siRNA treatment could signi cantly decrease the expression levels of NF-κB and p-NF-κB in APP/PS1 mice ( Fig. 8J and K, P < 0.01). These results suggested that NLRP1 knockdown could signi cantly improve the abnormal behavior and ameliorate the neuroin ammation in APP/PS1 mice.

NLRP1-siRNA treatment alleviates neuronal damage in APP/PS1 mice
We further used the H&E and Nissl staining to observe the effect of NLRP1-siRNA treatment on neuronal damage in APP/PS1 mice. The H&E results showed that there was signi cant neuronal damage and amyloid plaque in the cortex and hippocampus in NS and LV-scramble control group (Fig. 9A). Compared with the NS control group, NLRP1-siRNA treatment could signi cantly alleviate the neuronal damage and amyloid plaque in the cortex and hippocampus (Fig. 9A). The Nissl staining results also showed that NLRP1-siRNA treatment could signi cantly increase the Nissl bodies in the cortex and hippocampal CA3 regions compared with NS control group ( Fig. 9B and C, P < 0.05). To con rm the protective effect of NLRP1-siRNA treatment on neuronal damage in APP/PS1 mice, we further detected the expression levels of β-Gal and PSD95 by immunoblot. The results showed that NLRP1-siRNA treatment could signi cantly decrease the expression of β-Gal which was signi cantly increased in NS-and LV-scramble-treated APP/PS1 mice ( Fig. 9D and E, P < 0.01). And NLRP1-siRNA treatment signi cantly increased the expression of PSD95 compared with NS or LV-scramble control group (Fig. 9F and G, P < 0.01). These results indicated that NLRP1 knockdown could alleviate neuronal damage and synaptic dysfunction in APP/PS1 mice. NLRP1-siRNA treatment reduces Aβ deposition in APP/PS1 mice To con rm whether NLRP1 knockdown could ameliorate the Aβ generation and deposition, we further detected the protein and mRNA of APP and its metabolic enzymes by immunoblot and q-PCR. The immunoblot results showed that, compared with NS group, LV-scramble treatment had no effect on the APP, CTF-β, BACE and NCSTN expressions, while NLRP1-siRNA treatment could signi cantly decrease the APP, CTF-β and BACE expressions ( Fig. 10A-D, P < 0.01), and had no effect on NCSTN expression ( Fig. 9A and E). The q-PCR results also showed a signi cant decrease in APP and BACE mRNA expression in NLRP1-siRNA treatment group ( Fig. 10F and G, P < 0.01), and did not affect NCSTN mRNA expression (Fig. 10H).
Additionally, we further detected the expression of Aβ 1-42 by immuno uorescence and immunoblot. The results showed that, compared with the NS group, NLRP1-siRNA treatment signi cantly decreased the expression of Aβ 1-42 in the cortex and hippocampus in APP/PS1 mice (Fig. 10I-L, P < 0.05 or P < 0.01).
Meanwhile, we also detected the effect of NLRP1-siRNA treatment on Tau phosphorylation. The results showed that NLRP1-siRNA treatment could signi cantly decrease the level of p-TAU which signi cantly increased in the NS group ( Fig. 10M and N, P < 0.01). These results suggested that NLRP1 knockdown could inhibit Aβ generation and deposition in APP/PS1 mice.

Discussion
Alzheimer's disease (AD) is characterized by progressive cognitive impairments, Aβ deposition, Tau hyperphosphorylation and neurodegeneration in cortex and hippocampus. At present, it is still not completely understood the pathogenesis of AD, but growing evidence has focused on ROS oxidative stress and neuroin ammation, which may be closely involved in the pathogenesis in AD. Currently, there were no fully effective medicines and methods for the treatment of AD. Ginsenoside Rg1 (Rg1) is a bioactive component in ginseng, which has been used to delay senescence since ancient times. Our previous studies suggested that Rg1 could ameliorate H 2 O 2 -induced hippocampal neuron damage in vitro [12]. The APP/PS1 transgenic mouse is the most commonly used AD model animal, which exhibits obvious elevation of Aβ deposition and behavioral abnormalities. In this study, we hypothesized that Rg1 treatment might improve learning and memory, alleviate Aβ deposition and neuronal damage by inhibiting NOX2-NLRP1 in ammasome activation and reducing ROS oxidative stress and neuroin ammation in APP/PS1 mice. The present results showed that Rg1 treatment could signi cantly alleviate the learning and memory impairments and neuronal damage, reduce the expressions of NOX2 and NLRP1 in ammasome, inhibit the expressions of APP, BACE and Aβ 1-42 , and decrease Aβ deposition and Tau hyperphosphorylation in APP/PS1 mice. Meanwhile, the results also showed that apocynin (a NOX inhibitor) and NLRP1-siRNA treatment had similar effects to Rg1 in the APP/PS1 mice. Our results suggest that Rg1 treatment may be a potential therapeutic strategy for delaying the progression of AD.
The APP/PS1 transgenic mouse is the most commonly used AD model, which develops spatial memory impairment, increased Aβ deposition in the brain, synaptic loss, and neuronal loss similar to those features observed in AD [39,40]. It has been reported that the amyloid plaque can occur in the neocortex at 6-8 weeks and signi cantly increase at 8 months old [41], the synaptic dysfunction appears at about 3 months old [42], and the cognitive disability starts to show at about 6 months old in the APP/PS1 mice [43]. Therefore, we observed the effect of Rg1 treatment on APP/PS1 mice from 6M to 9M in the present study. In this study, the behavior results indicated that the APP/PS1-6M mice showed an increase of hyperactivity in the open eld test (OFT) which also reported in other studies [44] and slight impairments of learning and memory in Morris water maze (MWM) test. But the APP/PS1-9M mice showed signi cant hyperactivity and learning and memory impairments compared with WT-9M group. We speculated that excessive Aβ generation and deposition might excite brain inducing hyperactivity and neuronal damage in the APP/PS1 mice. A recent study suggests that hyperactivity is an anxiety symptom, which negatively moderates the effect of Aβ on cognitive decline in preclinical AD, resulting in a more rapid decline in several cognitive domains [45]. The present results showed that Rg1 treatment could signi cantly decrease the hyperactivity and alleviate learning and memory impairments, suggesting a protective effect of Rg1 on abnormal behavior in APP/PS1-9M mice. Additionally, it was found that olfactory dysfunction (OD) occurred in AD patients, even in those who were at an early stage of AD [46]. A clinical study also suggests that OD is prevalent in AD patients, which shows the overall declines of the olfactory threshold, discrimination and identi cation [29]. The present results of the buried pellet test showed that there was signi cant OD in the APP/PS1-9M group mice which the latency to nd food in the buried pellet test was signi cantly increased. However, treatment with Rg1 signi cantly reduced the latency to nd food in the buried pellet test, suggesting Rg1 improved the OD in APP/PS1-9M mice.
AD-induced behavioral dysfunction is thought to be related to the structural and functional impairments of speci c brain regions [47]. The cortex and hippocampus are primarily vulnerable regions affected by Aβ [48]. In this study, we found that there were slight Aβ deposition and neuronal damage in the cortex and hippocampus in APP/PS1-6M mice, while in the APP/PS1-9M mice, there were signi cant neuronal damage and Aβ deposition, and the Nissl bodies were signi cantly decreased in the cortex and hippocampus. However, Rg1 treatment could signi cantly alleviate neuronal damage and Aβ deposition.
To con rm the protective effect of Rg1 treatment on APP/PS1 mice, we further examined the expression of senescence-associated β-Gal, which is a critical biomarker of cell aging. The results showed that Rg1 treatment also signi cantly decreased the expression of β-Gal, which was signi cantly increased in APP/PS1-9M mice. The PSD95 is crucial for synaptic plasticity and is widely accepted as a regulatory protein that can regulate synapse function [49]. It has been reported that PSD95 is signi cantly reduced in the cortex and hippocampus of AD patients and negatively correlated with the severity of cognitive de cits [50]. The present study indicated that the expression of PSD95 was signi cantly decreased in the cortex and hippocampus, and was signi cantly increased after Rg1 treatment for 12 weeks in APPP/PS1-9M mice, suggesting that Rg1 treatment might reverse the synaptic dysfunction of AD mice. These data suggested that the changes in neuronal damage and Aβ deposition were along with behavioral disorders in APP/PS1 mice, and Rg1 treatment could signi cantly alleviate the Aβ deposition and neuronal damage in APP/PS1 mice.
Amyloid plaques are important pathological characteristics of AD. Aβ, the main component of amyloid plaques, was shown to play a crucial role in the development of AD pathologies [51]. Recent study reported that excessive Aβ generation and deposition could change cell metabolism and trigger Tau hyperphosphorylation, synaptic dysfunction, oxidative stress and neuroin ammation, further leading to progressive recognition dysfunction [1]. Aβ is produced by sequential cleavage of β-and γ-secretases at the C terminus of amyloid precursor protein (APP) [51]. The APP can be cleaved by α-, β-and γ-secretase into multiple peptides and fragments complex. APP can be cleaved by β/γ-secretase into pernicious Aβ peptide and SAPPβ fragments [52]. And APP can also be cleaved by α/γ-secretase to produce neuroprotective fragments of SAPPα and a nonamyloidogenic peptide [53]. BACE (beta-site APP-cleaving enzyme) is regarded as possessing β-secretase activity. Elevating the levels of BACE accelerates cleavage of APP into CTF-β. NCSTN is one of proteins that is regarded as possessing γ-secretase activity, which further cleaves CTF-β into Aβ. In the present study, the results showed that Rg1 treatment had no signi cant effect on NCSTN expression, but could signi cantly decrease the expression levels of APP, BACE and CTF-β, which were signi cantly increased in APP/PS1-9M mice. Meanwhile, we also detected the expression of Aβ 1-42 , a pernicious peptide, and the levels of amyloid plaque in the cortex and hippocampus. The results showed that Rg1 treatment signi cantly decreased the expression of Aβ  and the levels of amyloid plaque in APP/PS1 mice. Additionally, Rg1 treatment signi cantly decreased the levels of p-Tau and senescence-associated β-Gal that were also signi cantly increased in the APP/PS-9M mice. These results suggest that Rg1 may not only decrease APP expression but also inhibit the βsecretase pathway in APP processing to hinder pernicious Aβ formation.
Although the Aβ cascade is the most important pathogenesis of AD, there are still no clinical effective drugs acting on Aβ at present, suggesting other pathogenesis may be involved in AD. Growing evidence suggests that ROS oxidative stress and neuroin ammation play important roles in the progression of AD [19,54]. Excessive ROS accumulation may even be one of the rst pathogenic events during the progression of AD [55]. The NADPH oxidase (NOX) is an important enzyme that is responsible for ROS generation in many neurodegenerative diseases [56]. The NOX complex is currently the only enzyme family known to produce ROS as its sole function and comprises several isoforms such as NOX1-5.
NOX2 is mainly expressed in neurons and composed of the membrane heterodimers of gp91phox (NOX2) and p22phox, and cytosolic subunits of p47phox, p67phox, p40phox and RAC [57]. NOX2-mediated oxidative stress is closely involved in the progression of AD [58]. Neuroin ammation is considered to be a reaction and contributor to the neuronal injury and also plays an important role in the pathogenesis of AD [59]. In ammasome plays a critical role in regulating the in ammatory response and contributes to the pathogenesis of various neurologic diseases [60]. The in ammasome is large intracellular multiprotein complexes, which contain a member of the NOD-like receptor (NLR) family, such as NLRP1-3 and IPAF, by which they are de ned. The NLR protein recruits the in ammasome adaptor protein ASC, which in turn interacts with procaspase-1 leading to its activation, resulting in the maturation of the proin ammatory cytokines IL-1β and IL-18. NLRP1 is the rst reported in ammasome and is expressed extensively in many tissues. High-level of NLRP1 was found to be expressed in the central nervous system, especially in neurons [61]. Inhibition of NLRP1 expression could ameliorate age-related cognitive de cits [22]. However, it is still unclear whether activation of NOX2-NLRP1 in ammasome is involved in Aβ generation and deposition in the progression of AD.
Growing evidence suggests that Rg1 exerts signi cant antioxidant, anti-in ammatory and protective effects in neurological disease models in vivo and in vitro studies [62,63]. Our previous study demonstrated that Rg1 treatment could reduce oxidative stress and neuroin ammation and delay neuronal senescence in hippocampal neurons in vitro [17]. The present results showed that Rg1 treatment signi cantly decreased the levels of ROS, IL-1β in cortex and hippocampus. The further results demonstrated that Rg1 could signi cantly decrease the NOX2-related protein and mRNA levels of NOX2, p22phox and p47phox, and NLRP1 in ammasome-related protein and mRNA levels of NLRP1, ASC, caspase-1 and IL-1β. Meanwhile, Rg1 treatment signi cantly decreased the level of p-NF-κB in the APP/PS1 mice. These data suggested that the protective effects of Rg1 on APP/PS1 mice were closely related to inhibition of the NOX2-NLRP1 in ammasome pathway.
To further understand the molecular mechanisms underlying Rg1 actions, we examined the effects of apocynin and NLRP1-siRNA treatment on APP/PS1 mice. Apocynin is considered as an inhibitor of NOX to inhibit the generation of ROS. It has been reported that apocynin inhibits the NOX2 expression and ROS generation which increased by ischemia/reperfusion and signi cantly reduces the expressions of the NLRP3 in ammasome, IL-1β and IL-18 in a mouse model of stroke [56]. NLRP1-siRNA has commonly used to down-regulate the expression of NLRP1 in ammasome. Our previous study indicated that NLRP1-siRNA (408) signi cantly reduced expressions of NLRP1 in ammasome and improve learning and memory impairments in dexamethasone-induced mice [26]. The present study showed that both apocynin and NLRP1-siRNA treatment could signi cantly alleviate the learning and memory impairments, synaptic dysfunction and neuronal damage in the APP/PS1 mice. The further results showed that apocynin and NLRP1-siRNA treatment could signi cantly decrease the expressions of APP, CTF-β, BACE and Aβ1-42, and increase PSD95 expression in the APP/PS1 mice. These data suggest that both apocynin and NLRP1-siRNA treatment have protective effects on APP/PS1 AD mice. Furthermore, the results also showed that apocynin treatment could signi cantly decrease ROS generation and downregulate the expression of NOX2-NLRP1 in ammasome in APP/PS1 mice. And NLRP1-siRNA treatment also signi cantly down-regulated the expression of NLRP1 in ammasome in the APP/PS1 mice. These data suggested that activation of NOX2-NLRP1 in ammasome was closely involved in the progression of AD and down-regulation of NOX2-NLRP1 pathway signi cantly protected against cognitive dysfunction, Aβ deposition and neuronal damage in the APP/PS1 AD mice.

Conclusions
In summary, the present study suggested that Rg1 treatment could alleviate learning and memory impairment, synaptic dysfunction and neuronal damage and reduced Aβ generation and deposition in APP/PS1 mice. Meanwhile, we further found that Rg1 could signi cantly decrease the levels of ROS production and IL-1β, and signi cantly inhibited the NOX2-NLRP1 in ammasome pathway in the brain cortex and hippocampus in APP/PS1 mice. These ndings provided support that Rg1 could improve neuronal damage and delay the progression of AD, the mechanisms may be related to reducing neuroin ammation due to inhibition of NOX2-NLRP1 in ammasome activation. However, this study only provided a basic experiment for Rg1 in delaying the progression of AD in vivo, and the precise mechanisms of Rg1 on the regulation of the NOX2-NLRP1 pathway warrant further investigation. (NCP); (J) The swimming time in the quadrant of platform (STP, s). Results are expressed as mean ± SD. APP/PS1-6M group, n = 9; orther groups, n = 12. *P < 0.05, **P < 0.01 vs WT-9M group; #P < 0.05, ##P < 0.01 vs APP/PS1-6M group; ΔP < 0.05, ΔΔP < 0.01 vs APP/PS1-9M group; &P < 0.05, &&P < 0.01 vs D1 of 6M group in OFT and orientation navigation test of MWM.  Results are expressed as mean ± SD. *P < 0.05, **P < 0.01 vs WT-9M group; #P < 0.05, ##P < 0.01 vs APP/PS1-6M group; ΔP < 0.05, ΔΔP < 0.01 vs APP/PS1-9M group.   (NCP); (J) The swimming time in the quadrant of platform (STP, s). Results are expressed as mean ± SD, n = 8. *P < 0.05, **P < 0.01 vs APP/PS1-9M group; &P < 0.05, &&P < 0.01 vs D1 of 6M group in OFT and orientation navigation experiment of MWM.