C57BL/6 male mice (25–30 g) (Beijing Vital River Laboratory Animal Technology Co., Ltd) were housed in an animal care facility at the Children’s Hospital of Chongqing Medical University under standard laboratory conditions, which included temperature and humidity control, individual housing in plastic cages, and a 12-h light/dark cycle. The mice were given ad libitum access to food and water. All animal experiments were conducted by the Chongqing Science and Technology Commission guidelines and approved by the Chongqing Medical University Animal Care Committee. Every effort was made to minimize animal suffering and the number of animals used.

Aβ_1–42 peptides (A9810, Sigma, USA) were initially dissolved in hexafluoroisopropanol (HFIP) (AG968, Sigma, USA) and left at room temperature for 60 min. The resulting peptide film was then dissolved in dimethyl sulfoxide (DMSO) (D2650, Sigma, USA) to achieve a concentration of 5 mM. The solution was diluted to 100 μM with phosphate-buffered saline (PBS) and incubated at 4°C for 48 h to form oligomers.

The lateral ventricular administration of Aβ_1–42 is a commonly used experimental method to replicate the pathophysiology of Alzheimer’s disease (AD) (Du et al., 2019). Following anesthesia with 60 mg/kg sodium pentobarbital (i.p.), the mice’s cranial hair was removed, and their heads were secured in a stereotactic device in a horizontal position. The scalp was disinfected with povidone-iodine solution, and a midline incision was made to expose the skull. 2.5 μL of Aβ_1–42 (100 μM) was administered at a rate of 0.5 μL/min, targeting coordinates −0.5 mm posterior, +1.1 mm lateral, and − 3.0 mm ventral relative to bregma. The syringe was left in place for 5 min to ensure complete diffusion of the fluid. After carefully removing the syringe and suturing the scalp, the mice were returned to their cages for rewarming.

Parishin A was (HP265836, Chenguang Biology, Baoji, Shanxi, China dissolved in sterile PBS). Animals were randomly assigned to different treatment groups: WT, WT + Aβ, and WT + Aβ + PA. The WT + Aβ + PA group received an intraperitoneal injection of PA (10 mg/kg) starting 1 week before the Aβ_1–42 microinjection and continuing until the completion of the behavioral tests. The WT and WT + Aβ groups were administered the same volume of PBS.

Mice received intraperitoneal injections of PA starting at 8 weeks of age, followed by lateral ventricular administration of Aβ_1–42 at 9 weeks. Behavioral assessments were conducted at 11 weeks, and sacrifice with sample collection occurred at 12 weeks and 2 days.

Three weeks after the intraperitoneal injection of PA, mice underwent the Novel Object Recognition (NOR) test. During the training phase, two identical, odorless, and fixed objects (A and B) were placed in the apparatus, approximately 10 cm away from each side wall. Each mouse was introduced into the apparatus at an equal distance from the objects, with its back facing them. Exploration behavior, defined as touching the object with the mouth or nose or approaching within 2–3 cm, was recorded using video equipment and software. The number of interactions, exploration time, and distance traveled around each object were measured over a 5-min period. Following a 2-h interval, the test phase commenced, in which one of the identical objects was replaced with a novel object. The mouse was reintroduced into the apparatus under the same conditions, and exploration behavior was recorded for 5 min. All experiments were tracked and analyzed using the ANY-maze tracking system (Stoelting, USA).

The Y-maze apparatus consisted of three opaque plastic arms, labeled A, B, and C, arranged at 120° angles to each other. Mice were placed at the center of the maze and allowed to explore for 5 min. Entry into an arm was counted when a mouse moved all four limbs into the arm. The sequence of arm entries, such as A-B-C or B-C-A, was recorded to calculate the alternation score. Between tests, the maze was cleaned with 75% alcohol, which was allowed to evaporate completely to eliminate odor interference. The experiments were documented using the ANY-maze tracking system from Stoelting, USA. This alternation score was determined using the formula: spontaneous alternation rate (%) = [(number of spontaneous alternations) / (total number of arm entries–2)] × 100.

Three weeks after the intraperitoneal injection of PA, mice were subjected to the Morris Water Maze task. The test was conducted in a circular pool with a diameter of 150 cm and a height of 50 cm, filled with water maintained at 21–22°C, made opaque with nontoxic white paint. Light blue curtains surrounded the pool to create an isolated environment, with three geometric shapes attached to the curtains as visual cues. A high-definition camera was positioned directly above the pool to record the animals’ movements. A platform with a diameter of 10 cm was placed in the middle of the third quadrant of the pool, with the water level kept 1 cm above the platform. The day before spatial training, the mice were allowed to swim freely in a pool without a platform for 120 s to acclimatize. Subsequently, the mice underwent acquisition training for five consecutive days, with four trials per day. In each trial, mice that failed to locate the platform within 120 s were gently guided to it and allowed to stay there for 20 s. On the day following the final training session, the platform was removed, and the mice underwent a 120-s probe test. All experiments were recorded using the ANY-maze tracking system (Stoelting, USA).

1.5% sodium pentobarbital 100 mL: Sodium pentobarbital 1.500 g was weighed precisely and placed into a beaker. Distilled water was then added to completely dissolve it and transferred into a 100 mL volumetric flask for constant volume. Sodium pentobarbital was prepared in a 1.5% solution of sterile saline at the usual dose of 30 mg/kg body weight, which translates to 0.2 mL per mouse. The administration of the drug was done gradually, especially more slowly after administering 3/4 of the planned dose, rather than all at once. During injection, the corneal reflex, muscle relaxation, and pain responses in the animals were closely monitored. Once adequate anesthesia was achieved, the drug injection was halted immediately.

Following the completion of behavioral experiments, the mice were euthanized using the previously described methods, and brain tissue was harvested as follows: The neck was severed with tissue scissors just behind the skull. The scalp was pulled aside, and a midline incision was made with a razor blade between the eyes. Thin scissors were inserted into the foramen magnum and used to cut laterally along the skull. Small incisions were then made from the midline toward the sides. Using forceps, the skull flaps were flipped toward the sagittal suture, and the procedure was repeated on the opposite side. The brain was bisected with a clean surgical blade.

To remove the hippocampus, two short microscopic tweezers were used. One tweezer was placed near the junction of the cerebellum and cortex, and the other was positioned at the same point to gently dissect the cortical hemisphere laterally, exposing the hippocampus. The brain was stabilized with one tweezer, while the other tweezer was moved slightly forward and laterally, applying careful pressure to the medial white matter tract of the hippocampus. The hippocampus was then scooped or rolled sideways onto filter paper with the smooth end facing up, and the procedure was repeated for the second hemisphere.

Mouse neuroblastoma cells stably expressing human Swedish APP 695 (N2A^APP), obtained from Professor Chunjiu Zhong (Fudan University, Shanghai, China) were cultured in a medium consisting of 90% Dulbecco’s modified Eagle’s medium (DMEM) (11960044, Gibco, USA), 10% fetal bovine serum (FBS) (10099, Gibco, USA), and 100 μg/mL of G418 (11811031, Gibco, USA) at 37°C in a 5% CO₂ atmosphere.

To detect the effects of PA on the PS1 degradation, N2A^APP cells were pretreated with 40 μM PA or a control solvent for 12 h. This was followed by the treatment of 100 μg/mL cycloheximide (CHX) (CST, Danvers, MA, United States) to inhibit protein synthesis for various time (0, 6, 12 and 24 h) (Du et al., 2019). To investigate whether PA affects the ubiquitin-proteasome or autophagy lysosome degradation pathways of PS1, N2A^APP cells were pretreated with either 50 μM CQ (HY-17589A, MCE, USA) or MG132 (HY13259, MCE, USA) for 1 h, followed by treatment with 40 μM Parishin A for 24 h.

Cells were seeded in 10-well plates at a density that allowed them to reach about 70% confluency by the following day. The human Tau plasmid was transfected into N2A cells. Forty-eight hours post-transfection, puromycin was added to the medium to select for successfully transfected cells by eliminating the non-transfected cells. Approximately 2 weeks later, cells were reseeded into 96-well plates at a density of one cell per well. The cells were then allowed to grow until they reached the desired confluency.

N2A^APP cells were seeded into 96-well plates and treated with gradient concentrations of PA (0, 20, 40, 80, 160, and 320 μM) for 24 h. Subsequently, 10 μL of Cell Counting Kit-8 (CCK-8) reagent (MCE, Shanghai, China) was added to each well and were incubated at 37°C for 3 h, protected from light. Cell viability was determined by measuring the absorbance at 450 nm using a microplate reader (Thermo Fisher Scientific, MA, USA). Each group was subjected to four independent experiments, with each experiment performed in triplicate.

Anti-C20 (1:1000) antibody used to detect APP and its β-CTFs was kindly provided by laboratory of Professor Weihong Song. Anti-BACE1 (1:1000, #ab183612), anti-PS1 (1:1000, #ab76083) antibodies were purchased from Abcam (Cambridge, MA, USA). Anti-Ubquitin (1:1000, #10201-2-AP), β-actin (1:3000, 60004-1-1 g) antibodies were obtained from Proteintech (Wuhan, Hubei, China), Anti-P62 (1:1000, H00008878-M01), anti-LC3 (1:1000, #12741) antibodies were purchased from Abnova (Taipei, Taiwan, China), anti- p-tau(Ser199)(1:1000AF2418), anti- p-tau(Ser396) (1:1000, AF3148), anti- p-tau(Ser404)(1:1000, AF31440), anti- p-tau(Thr181) (1:1000, AF31449) antibodies were purchased from affinity (Affinity Biosciences, OH, USA) Anti-Tau5 (1:1000, A23490) antibodies were purchased from abclonal (Wuhan, Hubei, China), Anti-SYP (1:1000, MAB1598) was purchased from CST (Danvers, Massachusetts, USA).

The cells were washed with ice-cold phosphate-buffered saline (PBS) and lysed on ice for 30 min in RIPA lysis buffer (Beyotime, Shanghai, China) containing protease inhibitors (Roche, Basel, Switzerland). After centrifugation at 12,000 rpm for 15 min at 4°C, the supernatant was collected. Protein concentration was determined using a BCA Protein Assay Kit (Thermo Fisher Scientific, MA, USA). Subsequently, 30 μg of total protein was denatured by boiling with 5× sample buffer at 95°C for 5 min. The samples were then separated using Tris-glycine SDS-PAGE gels (EpiZyme, Shanghai, China) and transferred to Immobilon-PTM polyvinylidene difluoride (PVDF) membranes (Millipore, MA, USA). The PVDF membrane was blocked with 5% bovine serum albumin (Sigma, St. Louis, MO, USA) for 1.5 h to minimize nonspecific binding. The membranes were then incubated with primary antibodies overnight at 4°C, followed by incubation with corresponding HRP-labeled secondary antibodies (goat anti-rabbit IgG or goat anti-mouse IgG, both at 1:3000 dilution, Perkin-Elmer) for 1–2 h at room temperature. The blots were visualized using the Bio-Rad Imager with Western ECL substrate (Bio-Rad, Hercules, CA, USA). Immunoblotting with GAPDH was performed to ensure equal loading and protein quality. Band intensities were quantified using Bio-Rad Quantity One software (Bio-Rad, Hercules, CA, USA).

Cells were lysed using IP buffer (Beyotime, Shanghai, China) supplemented with protease inhibitors for 30 min, and the lysates were then centrifuged at 12,000 g for 15 min at 4°C to obtain the supernatants. A total of 500 μg of protein from each sample was incubated overnight at 4°C with an anti-PS1 primary antibody or nonspecific IgG as a control. This was followed by an incubation with protein A/G magnetic beads for IP (Millipore, MA, USA) for an additional 2 h at 4°C. After incubation, the beads were washed four times with ice-cold PBS. The proteins bound to the beads were eluted by boiling in 2× SDS-PAGE loading buffer at 95°C for 5 min. These eluted proteins were then analyzed by subsequent immunoblotting.

Total RNA from cells and brain tissues were extracted using High Pure Total RNA Extraction Kit (Bio Teke, Peking, China) in accordence with the manufacturer’s protocol. The concentration and purity of RNA were assessed utilizing a NanoDrop 2000 spectrophotometer (Nanodrop Technologies, Wilmington, DE, USA). Subsequently, 1 μg of total RNA served as a template to synthesize single-stranded complementary DNA (cDNA) using the PrimeScript™ RT Reagent Kit (Takara, Otsu, Shiga, Japan). The synthesized cDNA was amplified by quantitative real-time PCR (qRT-PCR) using SYBR Premix Ex Taq II (Takara, Otsu, Shiga, Japan) and managed through CFX Manager software (Bio-Rad, Hercules, CA, USA). The primer used were as follows: PS1 (forward: 5′-GAGACTGGAACACAACCATAGCC, reverse: 5′-AGAACACGAGCCCGAAGGTGAT); GAPDH (forward: 5′-GGCATTGTGGAAGGGCTCAT, reverse: 5′-AGATCCACGACGGACACATT). IL-4 (forward: 5′- GGTCTCAACCCCCAGCTAGT, reverse: 5′- GCCGATGATCTCTCTCAAGTGAT); IL-6 (forward: 5′- TAGTCCTTCCTACCCCAATTTCC, reverse: 5′- TTGGTCCTTAGCCACTCCTTC). GAPDH was employed as an internal control for normalization, and the relative mRNA levels of PS1 were normalized to those of GAPDH. To calculate the ΔCt value for each sample, subtract the Ct value of the housekeeping gene GAPDH from the Ct value of the target gene. Then, compute the ΔΔCt value by subtracting the average ΔCt of the reference sample from the ΔCt of each individual sample. Finally, the relative expression of each gene is calculated using the formula 2^−ΔΔCt.

The brain tissues (cortex and hippocampus) from mice were homogenized to a 10% concentration in saline on ice, followed by centrifugation at 2500 rpm for 15 min at 4°C to collect the supernatant. The levels of pro-oxidants including hydrogen peroxide (H₂O₂), total nitric oxide synthase (T-NOS) and inducible nitric oxide synthase (i-NOS) as well as antioxidants including glutathione reductase (GR) and total antioxidant capacity (T-AOC) were measured using commercial assay kits (Jiancheng Biochemical, Nanjing, Jiangsu, China) according to the manufacturer’s instructions.

For the H₂O₂ assay, the supernatant was first mixed with reagents 1 and 2, incubated at 37°C for 1 min, followed by the addition of reagents 3 and 4. The mixtures were thoroughly mixed, and the absorbance at 405 nm was measured using a microplate reader.

For NOS detection, the supernatant was incubated with or without reagent 6 (for i-NOS and T-NOS, respectively), along with reagents 1, 2, and 3 at 37°C for 15 min. Reagents 4 and 5 were then added, mixed thoroughly, and the absorbance at 530 nm was recorded using a microplate reader.

For the GR assay, the supernatant was incubated with the working solution at 37°C for 30 s, and the initial absorbance (A1) was recorded at 340 nm. After a further incubation of 10 min at 37°C, a second absorbance (A2) was recorded. The level of GR was calculated as the difference between A2 and A1.

For T-AOC measurement, the supernatant was mixed with reagents 1, 2, and 3 and incubated at 37°C for 30 min. After adding reagents 4 and 5, the mixture was incubated for an additional 10 min at 37°C, and the final absorbance at 520 nm was recorded.

The mCherry-GFP-LC3B plasmid was utilized to establish a dual fluorescence autophagy system for assessing autophagy flux. In brief, cultured cells in a 6-well plate were transfected with the mCherry-GFP-LC3B plasmid using Lipofectamine 3000 (L3000008, Invitrogen, USA) for 24 h. After transfection, the cells were pretreated with PA (40 μM) for 6 h, and either with or without CQ (50 μM) for 24 h (Le et al., 2023; Yang et al., 2024). Subsequently, the cells were fixed with 4% paraformaldehyde (PFA) (158127, Sigma, USA), and the cell sheet was sealed with an anti-fluorescence quench agent to preserve fluorescence. Cellular images were captured using a Nikon 90i fluorescence microscope. Autophagy was quantified by counting GFP-LC3 and mCherry-LC3 dots in more than 15 cells from three replicate experiments.

The statistical analysis was performed using the SPSS software (Version 25.0), and the data were expressed as means ± SEM. Student’s t-test or one-way ANOVA was used to evaluate the results. The significance level was set at p < 0.05.

To evaluate the impact of PA on cell survival, N2A^APP cells were exposed to gradient concentrations of PA (ranging from 0 to 320 μM) for 24 h. Cell viability was then assessed using the CCK-8 assay. It was observed that all concentrations of PA had no significant effect on the viability of N2A^APP cells, except for the highest concentration of 320 μM, which resulted in a decrease in cell viability (20 μM: 70.79% ± 18.0.82%, p = 0.801 vs. 0 μM; 40 μM: 59.78% ± 7.49%, p = 0.222 vs. 0 μM; 80 μM: 86.70% ± 6.74%, p = 0.403 vs. 0 μM; 160 μM: 105.70% ± 7.81%, p = 0.715vs. 0 μM; 320 μM: 81.76% ± 13.70%, p = 0.257 vs. 0 μM; n = 4 in each group; Figure 1A).

Further, we evaluated the effect of PA on APP processing in N2A^APP cells. The findings revealed that treatment with PA at a concentration of 40 μM led to a reduction in PS1 expression (20 μM: 70.79% ± 18.82%, p = 0.081 vs. 0 μM; 40 μM: 59.83% ± 7.49%, p = 0.022 vs. 0 μM; 80 μM: 86.70% ± 6.74%, p = 0.040 vs. 0 μM; 160 μM: 105.70% ± 7.80%, p = 0.715 vs. 0 μM; 320 μM:81.76% ± 13.7%, p = 0.257vs. 0 μM; n = 3 in each group; Figures 1B,C). However, PA had no obvious impact on APP expression across all tested concentrations (20 μM: 104.80% ± 5.93%, p = 830 vs. 0 μM; 40 μM: 103.10% ± 8.98%, p = 0.898 vs. 0 μM; 80 μM: 110.10% ± 17.11%, p = 0.653 vs. 0 μM; 160 μM: 100.90% ± 20.08%, p = 0.969 vs. 0 μM; 320 μM: 119.50% ± 24.96%, p = 0.289 vs. 0 μM; n = 3 in each group; Figures 1B,D) and BACE1 (20 μM: 115.90% ± 10.31%, p = 0.462 vs. 0 μM; 40 μM: 108.90% ± 15.21%, p = 0.676 vs. 0 μM; 80 μM: 113.90% ± 14.12%, p = 0.519 vs. 0 μM; 160 μM: 116.34% ± 23.04%, p = 0.450 vs. 0 μM; 320 μM: 109.54% ± 15.40%, p = 0.657 vs. 0 μM; n = 3 in each group; Figures 1B,E). These data suggest that PA alleviates APP processing by attenuating PS1 expression.

Give that protein homeostasis is maintained by a balance between protein synthesis and degradation, we examined whether PA treatment affects either the synthesis or degradation of PS1, thereby influencing its expression levels. To determine the effects of PA on PS1 synthesis, qRT-PCR was utilized to measure PS1 mRNA levels in N2A^APP cells following PA treatment. It was observed that PA treatment did not alter PS1 mRNA levels in N2A^APP cells (n = 6, 92.45% ± 7.630%, p = 0.290 vs. 0 μM; Figure 2A).

To examine the impact of PA on the degradation of PS1, a protein synthesis inhibitor, cycloheximide (CHX), was applied to N2A^APP cells with or without PA treatment. The results indicated that PA treatment accelerated PS1 degradation compared to controls (p = 0.035 vs. CTR; n = 5 in each group; Figures 2B,C).

Eukaryotes have two major protein degradation systems, the ubiquitin-proteasome system (UPS) and the autophagolysosome pathway (ALP). To elucidate the potential involvement of the UPS in the observed reduction in APP processing, we employed MG132 to inhibit the proteasome. It was observed no significant change in PS1 ubiquitination levels (n = 4 in each group; Figure 2D). To test whether PA influences the degradation of PS1, chloroquine (CQ), an autophagy inhibitor, and MG132, a proteasome inhibitor, were used. CQ treatment resulted in an increase in PS1 expression, whereas MG132 had no effect on PS1 levels (n = 5, MG132: 124.10% ± 22.58%, p = 0.626 vs. CTR, CTR + CQ: 329.10% ± 54.44%, p = 0.001 vs. CTR in each group; Figures 2E,F).

To assess the effect of PA on autophagy, we examine the autophagy activity in N2A^APP cells and compared it with N2A cells. It was found that an increase in autophagic markers, such as P62 (n = 4, 221.10% ± 25.90%, p = 0.005 vs. N2A; Figures 3A,B) and LC3 II/I (n = 4, 173.20 ± 25.21%, p = 0.027 vs. N2A; Figures 3A,C) in N2A^APP cells, indicating enhanced autophagy accumulation in Alzheimer’s disease. Furthermore, we utilized a fluorescence assay using mCherry-GFP-LC3 to examine the autophagic flux in N2A^APP cells. This assay capitalizes on the differential stability of mCherry and GFP fluorescence in acidic environments; mCherry fluorescence persists, whereas GFP fluorescence is quickly quenched. Therefore, an increase in autophagic flux would elevate both yellow (combined mCherry and GFP) and red (only mCherry) puncta, whereas a blockage in autophagosome-lysosome fusion or a rise in lysosomal pH would predominantly increase yellow puncta. Our findings demonstrated a marginal increase in yellow dots (n = 16–25, 2.01 ± 0.01, p = 0.053 vs. N2A; Figure 3D) and a significant increase in red dots (n = 16–25, 2.50 ± 0.43, p = 0.006 vs. N2A; Figure 3E), suggesting a blockage in the fusion of autophagosomes with lysosomes or inhibition of lysosomal degradation. These findings indicate an accumulation of autophagy in AD.

To explore whether PA affects the autophagolysosome pathway, we treated N2A^APP cells with PA and assessed the levels of autophagy-associated proteins P62 and LC3. The results revealed that PA treatment led to a decrease in P62 (n = 6, 76.69% ± 3.64%, p = 0.001 vs. CTR; Figures 4A,B) levels and LC3 II/I ratios (n = 5, 75.19% ± 6.91%, p = 0.072 vs. CTR; Figures 4A,C), suggesting that PA may enhance autophagy function in AD model cells.

To further explore PA’s effects on the autophagolysosome pathway, we used CQ to inhibit this pathway and examined the expression of P62 and LC3 in N2A^APP cells treated with PA and CQ. The results showed that PA treatment resulted in an increase in LC3 II (n = 5, 165.70% ± 28.4%, p = 0.049 vs. CTR + CQ; Figures 4D,G) in N2A^APP cells. However, PA did not significantly alter P62 expression (n = 5, 96.93% ± 17.04%, p = 0.861 vs. CTR + CQ; Figures 4D,E) and LC3 II/I ratios (n = 5, 163.10% ± 72.71%, p = 0.410 vs. CTR + CQ; Figures 4D,F), suggesting that PA may counteract the autophagy inhibition caused by CQ.

Further analysis using a fluorescence assay with mCherry-GFP-LC3 revealed that PA treatment enhanced autophagic flux, as evidenced by an increase in both total (n = 11, PA: 17.40 ± 0.58, p = 0.001 vs. CTR, n = 17, PA + CQ: 39.41 ± 3.15, p = 0.008 vs. CTR + CQ in each group; Figures 4H,I) and red puncta (n = 12, PA: 9.44 ± 2.67, p = 0.178 vs. CTR, n = 21, PA + CQ: 14.50 ± 2.01, p = 0.030 vs. CTR + CQ in each group; Figures 4H,I), without affecting the yellow puncta (n = 12, PA: 7.67 ± 2.67, p = 0.533 vs. CTR, n = 21, PA + CQ: 23.63 ± 2.00, p = 0.331vs. CTR + CQ in each group; Figures 4H,I). In contrast, CQ treatment led to an increase in both total and yellow puncta while slightly reducing red puncta (n = 12, CTR + CQ: 10.00 ± 2.23, p = 0.064 vs. CTR). Notably, adding PA to CQ-treated cells resulted in an increase in total puncta, emphasizing PA’s potential to reverse CQ’s inhibitory effects on autophagy.

Given PA’s role in inhibiting APP processing, we explored its impact on the cognitive functions of AD mouse models. A mouse model was created through microinjection of Aβ_1–42 into the lateral ventricle of WT mice. We then assessed the therapeutic effects of PA through daily intraperitoneal injections of 10 mg/kg, starting 1 week before and continuing until after the behavioral tests. Two weeks post-Aβ_1–42 injection, the mice underwent the Morris water maze test to assess their spatial learning and memory (Figure 5A).

During the adaptation period, the average swimming distances were comparable across all groups (WT: 21.39 ± 0.68 m; WT + Aβ: 19.90 ± 0.92 m, p = 0.175 vs. WT; WT + Aβ + PA: 19.50 ± 0.59 m, p = 0.061 vs. WT + Aβ, p = 0.692 vs. WT; n = 12–18 in each group; Figures 5B,C), indicating that neither Aβ_1–42 nor PA treatment affected motor function. During the training phase, mice treated with Aβ_1–42 showed significant spatial learning deficits, with increased escape latency to locate the hidden platform compared to WT mice. However, PA treatment substantially reduced the escape latency in Aβ_1–42 treated mice (p < 0.001 vs. WT + Aβ; p = 0.845 vs. WT; n = 12–18 in each group; Figure 5D). In the probe test, which measures memory retrieval, Aβ_1–42 treated mice demonstrated significantly a delayed latency to first entry into platform-zone (WT: 10.28 ± 1.27 s; WT + Aβ: 60.18 ± 12.82 s, p < 0.001 vs. WT; n = 12–18 in each group; Figure 5E) and fewer platform-zone crossing (WT: 4.308 ± 0.58; WT + Aβ: 1.58 ± 0.36, p = 0.001 vs. WT; n = 12–18 in each group; Figure 5F). PA treatment markedly improved these metrics, restoring them to levels similar to the WT group (for first entry into platform-zone: WT + Aβ + PA: 11.41 ± 2.15 s, p < 0.001 vs. WT + Aβ, p = 0.900 vs. WT, n = 12–18 in each group, Figure 4E; for platform-zone crossing: WT + Aβ + PA: 4.22 ± 0.38, p = 0.001 vs. WT + Aβ, p = 0.890 vs. WT, n = 12–18 in each group; Figure 5F).

To further investigate the impact of PA on cognition. Novel object recognition test and Y maze tests were conducted. The findings indicated that the Aβ_1–42 significantly reduced the recognition index and spontaneous alteration rate (For novel object recognition test: WT: 56.54 ± 2.57; WT + Aβ: 46.63 ± 1.85, p = 0.043 vs. WT; n = 10 in each group; Figures 5G,H; For Y maze: WT: 48.49 ± 1.70; WT + Aβ: 40.31 ± 3.00, p = 0.028 vs. WT; n = 10 in each group; Figures 5I,J). However, PA treatment markedly improved these measures, restoring them to levels observed in the WT group (For novel object recognition test: WT + Aβ + PA: 22.80 ± 2.77, p < 0.001 vs. WT + Aβ; n = 10 in each group; Figures 5I,J; WT + Aβ: 64.58 ± 4.75, p < 0.001 vs. WT + Aβ; n = 10 in each group; Figures 5G,H; For Y maze: WT + Aβ + PA: 48.88 ± 2.60, p = 0.022 vs. WT + Aβ; n = 10 in each group; Figures 5I,J). These results suggest that PA effectively counters cognitive decline in AD model mice.

To explore the effect of sex on AD, we counted the cognitive coefficients and spontaneous alterations of female and male mice separately for each group and found no significant difference between the sexes (For novel object recognition test: male: WT: 54.64 ± 4.12; WT + Aβ: 48.38 ± 5.31, p = 0.823vs. WT; WT + Aβ + PA: 64.64 ± 12.26, p = 0.076 vs. WT + Aβ; female: WT: 59.42 ± 5.13; WT + Aβ: 44.88 ± 4.51, p = 0.128 vs. WT; WT + Aβ + PA: 64.52 ± 19.64, p = 0.052 vs. WT + Aβ; n = 10 in each group; Figure 5H; For Y maze: male: WT: 45.31 ± 9.96; WT + Aβ: 35.34 ± 2.34, p = 0.126 vs. WT; WT + Aβ + PA: 47.65 ± 12.31, p = 0.053 vs. WT + Aβ; female: WT: 51.66 ± 6.39; WT + Aβ: 45.27 ± 1.55, p = 0.460 vs. WT; WT + Aβ + PA: 50.11 ± 4.84, p = 0.674 vs. WT + Aβ; n = 10 in each group; Figure 5L).

Oxidative stress is a significant pathological consequence of Aβ pathology and plays a crucial role in the pathogenesis of AD. We therefore investigated the potential role of PA in oxidative stress within an Aβ_1–42-induced AD mouse model. Our results indicated significant elevations in pro-oxidants levels including hydrogen peroxide (H₂O₂) (134.00% ± 6.91%, p < 0.01 vs. WT; n = 5–8 in each group; Figure 6A) and inducible nitric oxide synthase (i-NOS) (165.8% ± 7.08%, p < 0.001 vs. WT; n = 8 in each group; Figure 6B) in mice treated with Aβ_1–42. In contrast, antioxidants levels, such as glutathione reductase (GR) (64.13% ± 6.53%, p = 0.038 vs. WT; n = 11–13 in each group; Figure 6D) and total antioxidant capacity (T-AOC) (69.68% ± 12.11%, p < 0.033 vs. WT; n = 7–8 in each group; Figure 6E) were significantly decreased in these mice. Notably, PA treatment effectively normalized both the increased pro-oxidants and the reduced antioxidants levels (for H₂O₂: 99.77% ± 8.94%, p < 0.001 vs. WT + Aβ, p = 0.985vs. WT, n = 5–8 in each group, Figure 6A; for i-NOS: 114.6% ± 9.08%, p = 0.001vs. WT + Aβ, p = 0.197 vs. WT, n = 8 in each group; Figure 6B; for GR: 104.8% ± 12.65%, p = 0.017 vs. WT + Aβ, p = 0.771 vs. WT, n = 11–13 in each group; Figure 6D; for T-AOC: 111.5% ± 7.01%, p < 0.013 vs. WT + Aβ, p = 0.661 vs. WT, n = 7–8 in each group; Figure 6E). However, neither Aβ_1–42 nor PA treatment had an impact on the levels of total nitric oxide synthase (T-NOS) (WT + Aβ: 94.16% ± 10.77%, p = 0.770 vs. WT; WT + Aβ + PA: 97.73% ± 13.63%, p = 0.858 vs. WT + Aβ, p = 0.909vs. WT; n = 8 in each group; Figure 6C). Thus, these results affirm the anti-oxidative properties of PA in reducing oxidative stress in AD model mice.

We also explored the anti-inflammatory response of PA in a mouse model of Aβ_1–42 induced AD. Using qRT-PCR analysis, we observed that interleukin 4 (IL-4), an anti-inflammatory cytokine, was significantly reduced to 69.14% ± 6.01% (p = 0.04 vs. WT; n = 5 per group; Figure 6F) compared to the AD group. Pro-inflammatory cytokines such as tumor necrosis factor-alpha (TNF-α) and interleukin 6 (IL-6) were notably increased in the AD group, with TNF-α reaching 257.50% ± 35.73% (p < 0.001 vs. WT; n = 5 per group; Figure 6G) and IL-6 at 145.20% ± 22.01% (p = 0.032 vs. WT; n = 5 per group; Figure 6H). Notably, PA treatment significantly elevated IL-4 levels to 149.40% ± 12.85% (p < 0.001 vs. WT + Aβ + PA; n = 5 per group; Figure 6F) and decreased the levels of pro-inflammatory cytokines, with TNF-α reduced to 137.80% ± 23.50% (p = 0.006 vs. WT + Aβ + PA; n = 5 per group; Figure 6G) and IL-6 to 91.80% ± 7.88% (p = 0.028 vs. WT + Aβ + PA; n = 5 per group; Figure 6H), highlighting its potential therapeutic benefits in reducing inflammation in AD.

To assess whether PA could alleviate synaptic dysfunction in AD, we examined the levels of synaptic-related proteins, including postsynaptic density protein 95 (PSD95) and synaptic vesicle protein (SYP). Our findings showed that while there was no significant change in PSD95 levels among the groups (n = 6, WT + Aβ: 86.98% ± 15.78%, p = 0.408 vs. WT: 89.5% ± 9.87%, p = 0.869 vs. WT + Aβ + PA; Figures 6I,J), SYP levels were notably reduced in Aβ-treated mice. However, PA treatment was able to rescue SYP levels (n = 6, WT + Aβ: 54.7% ± 7.10%, p = 0.071 vs. WT: 112.10% ± 27.07%, p = 0.027vs. WT + Aβ + PA; Figures 6I,K), suggesting that PA has the potential to improve synaptic dysfunction in AD.

We explore whether PA affects both total Tau and phosphorylated Tau expression. In mice induced with Aβ, no significant changes were observed in levels of total Tau (n = 6, WT + Aβ: 152.80% ± 33.64%, p = 0.116 vs. WT: 119.80% ± 18.64%, p = 0.309 vs. WT + Aβ + PA; Figures 7A,B) or phosphorylated Tau at various sites (Ser396: n = 6, WT + Aβ: 123.50% ± 25.54%, p = 0.497 vs. WT, 145.30% ± 32.49%, p = 0.528 vs. WT + Aβ + PA; Figures 7A,C, Thr181: WT + Aβ: 152.70% ± 37.63%, p = 0.199 vs. WT, 136.60% ± 29.08%, p = 0.631 vs. WT + Aβ + PA; Figures 7A,D, Ser404: WT + Aβ: 81.54% ± 24.03%, p = 0.697 vs. WT, 165.60% ± 51.51%, p = 0.091 vs. WT + Aβ + PA; Figures 7A,E, Ser199: WT + Aβ: 156.70% ± 66.57%, p = 0.436 vs. WT, 190.60% ± 55.74%, p = 0.640 vs. WT + Aβ + PA; Figures 7A,F). This may be due to the inability of endogenous Tau in mice to form neurofibrillary tangles.

To further explore PA’s effect on human Tau, we utilized stabilized N2A cells transfected with human Tau, which successfully and stably overexpressed tau5 (n = 3, N2A^Tau: 301.50% ± 53.18%, p = 0.019 vs. N2A; Figures 7G,H). The results demonstrated that PA treatment significantly reduced the expression of Tau5 (n = 5, N2A^Tau: 90.63% ± 10.55%, p = 0.002 vs. N2Atau + PA; Figures 7I,J) and phosphorylated Tau at Ser199 (n = 5, N2A^TTau: 69.89% ± 0.07%, p = 0.017 vs. N2A^Tau + PA; Figures 7I,K). However, no significant changes were noted in the phosphorylation levels at Ser396 (n = 5, N2A^Tau: 84.41% ± 21.99%, p = 0.499 vs. N2A^Tau + PA; Figures 7I,L) and Ser404 (n = 5, N2A^Tau: 101.20% ± 20.29%, p = 0.954 vs. N2A^Tau + PA; Figures 7I,M). These data suggest that PA has a mitigating effect on the levels of both total Tau and phosphorylated Tau at various phosphorylation sites.