Male C57BL/6 mice (15 weeks old) weighing approximately 23–28 g were purchased from Jiangsu Jicui Pharmachem Biotechnology Co., Ltd. (Nanjing, Jiangsu, China). All mice were housed (five animals per cage) in specific pathogen-free facilities at the Guangdong Medical University Laboratory Animal Center. They were maintained at a constant temperature of 23 $\pm$ 2 °C, a relative humidity of 60 $\pm$ 5%, a 12-h light/12-h dark cycle, and with free access to food and water. Bedding and fresh water were replaced weekly, and every 5 days post-modeling. All animal-related experiments in this study were approved by the Ethics Committee of Guangdong Medical University (Ethics Approval No. GDMU-2023-000109) and adhered to international practices for animal experiments.

Phosphate-buffered saline (PBS), empty adeno-associated virus (AAV), and draxin-expressing AAV (GOSV0403972; Shanghai Genechem Co., Ltd., Shanghai, China) were delivered into the hippocampus of 15-week-old male C57BL/6 mice using stereotactic techniques. The animals were securely positioned in a stereotactic frame (018.116; NEUROSTAR, Malvern, PA, USA) with the head maintained in a flat skull position. The anterior fontanelle was exposed following a midline scalp incision with ophthalmic scissors. According to the mouse brain stereotactic atlas, the Cornu Ammonis 1 (CA1) region was targeted with the following coordinates relative to the anterior fontanelle: 2.1 mm posterior, 1.8 mm lateral from the midline, and 2.0 mm ventral from the skull surface. A total volume of 1 µL of each solution was slowly infused using a micro-syringe (980-02918-00; Ningbo Zhenhai Glass Instrument Factory, Ningbo, Zhejiang, China) at a rate of 0.1 µL/min. The needle was kept in place for an additional 5 min to facilitate adequate diffusion of the solution and minimize backflow. Subsequently, the needle was gradually withdrawn to prevent reflux. The wild-type C57BL/6 mice were assigned to three experimental groups based on the injected material. The wild-type (WT) +PBS group was injected with PBS, the WT+AAV-NC (negative control) group was injected with empty AAV, and the WT+AAV-draxin group was injected with draxin-expressing AAV. At the end of all experimental procedures (including behavioral tests), mice were euthanized via an intraperitoneal injection of sodium pentobarbital (150 mg/kg, Sigma-Aldrich, St. Louis, MO, USA) to ensure a humane death. Euthanasia was confirmed by the absence of vital signs (cessation of breathing, loss of corneal reflex, and no response to paw pinch) for at least 5 min. The brains were then rapidly dissected for subsequent histological and molecular analyses.

The HT22 ( iCell-m020) mouse hippocampal neuronal cell line used in the experiments was purchased from Cellverse (iCell) Bioscience Technology Co., Ltd. (Shanghai, China). All cells were authenticated by the vendor and tested negative for mycoplasma prior to use. Cell experiments were conducted in a sterile, ultra-clean facility within the cell room of the Institute of Neurology at Guangdong Medical University, adhering strictly to aseptic requirements. The cells were cultured in an incubator under optimal conditions, featuring a gas phase composition of 95% air and 5% carbon dioxide, a constant temperature of 37 °C, and a humidity range of 70%–80%.

(1) Lentivirus: pHBLV-CMV-MCS-3FLAG-EF1-ZsGreen-T2A-PURO, an empty load virus, and a draxin-overexpressing virus were purchased from Hanheng Biotechnology Co., Ltd. (HH20221222GYQ-LV01, Shanghai, China).

(2) Adeno-associated virus: GV753-hsyn promoter-EGFP-FT2A-MCS-WPRE-SV40 PolyA, empty load virus, and draxin-overexpressing virus were purchased from GeneChem Co., Ltd. (GOSV0403972, Shanghai, China).

Cells with favorable viability were plated into 96-well plates for subsequent experiments, with cell numbers adjusted for the specific experimental requirements. For cell viability assays, approximately 2 $\times$ 10³ cells were seeded per well, and the assay was performed after cells had fully adhered to the plate and reached an appropriate density. To determine the optimal A$\beta$42 (04010011526, Qiangyao Biotech Co., Ltd., Shanghai, China) concentration, around 1 $\times$ 10⁴ cells were seeded into each well, followed by incubation with A$\beta$42 for 24 h before the assay. The CCK8 (CK04-11, DOJINDO Laboratories, Kyushu, Japan) assay was performed according to the manufacturer’s instructions.

Tissues were homogenized in lysis buffer (2500070007, Solarbio, Beijing, China) containing protease and phosphatase inhibitors (A32961; Thermo Fisher Scientific, MA, USA) and centrifuged at 12,000 rpm for 20 min at 4 °C. Supernatant was collected and the protein concentrations quantified using a bicinchoninic acid assay (BCA) (3213970, ThermoFisher Scientific, Waltham, MA, USA). Proteins (45 µg) were mixed with loading buffer, electrophoresed on SDS-PAGE gel, and then transferred to a polyvinylidene difluoride membrane (PVDF IPVH00010, Millipore Corporation, Burlington, MA, USA). The membranes were blocked with TBS/0.1% Tween-20 (TBST) buffer containing 5% bovine serum albumin (V900933-100G; Sigma-Aldrich) at room temperature, then incubated overnight at 4 °C with MAP2 antibody (rabbit, 1:1000, 8707S, Cell Signaling Technology, Danvers, MA, USA), NeuN antibody (rabbit, 1:1000, 24307S, Cell Signaling Technology), phosphorylated tau (Ser396) (p-tau396) antibody (rabbit, 1:1000, AB109390, Abcam, Cambridge, Cambridgeshire, UK), phosphorylated tau (Thr231) (mouse, 1:1000, PA5-117230, Invitrogen, MA, USA), draxin (sheep, 1:1000, CDUB03240, R&D Systems, Minneapolis, MN, USA), or primary antibodies to $\beta$-catenin (1:1000, 8480, Cell Signaling Technology), GSK-3$\beta$ (1:1000, 9315S; Cell Signaling Technology), and lipoprotein receptor-related protein 6 (LRP6) (1:1000, 3395S; Cell Signaling Technology). All primary antibodies were obtained from Cell Signaling Technology. The membranes were then washed with TBST (five times for 3 min each) and incubated with secondary antibodies (1:10,000 in TBST, BA1054, Boster Biological Technology Co., Ltd., Wuhan, Hubei, China) for 1 h. Signals were visualized using enhanced chemiluminescence (ECL) (BMU102-CN, Abbkine Scientific Co., Ltd., Wuhan, Hubei, China) (Azure Biosystems C600, Dublin, CA, USA) and optical densities were analyzed using ImageJ software (v1.54f, National Institutes of Health, Bethesda, MD, USA). Specifically, quantification software and settings were as follows: the “Gels” tool of ImageJ v1.54f was used and band optical density (OD) was measured after background subtraction (rolling ball radius = 50 pixels). Additionally, normalization details were verified: loading control bands were confirmed to be within the linear range of detection via serial dilution validation, ensuring accurate normalization. For averaging of replicates, data from 3–4 independent biological replicates were first normalized to the loading control [vinculin (1:1000, 13901S; Cell Signaling Technology) or (glyceraldehyde-3-phosphate dehydrogenase) GAPDH (1:1000, 5174S; Cell Signaling Technology)] for each sample, then averaged to calculate the relative expression level [mean $\pm$ standard error of the mean (SEM)]. Furthermore, for western blot (WB) densitometry statistical analysis, one-way ANOVA (for multiple groups) or two-way ANOVA (for factorial designs) was used to compare relative expression levels, with Tukey’s post-hoc test for pairwise comparisons (significance set at p 0.05). Original Western blot images are provided in Supplementary Material-WB Original.

Relative mRNA levels were measured via quantitative real-time polymerase chain reaction (qRT-PCR) with validated primer pairs (Table 1). Total RNA was extracted from cells using TRIzol reagent (15596018CN, Ambion, Austin, TX, USA) and cDNA was synthesized using a reverse transcription kit (RR820Q, TaKaRa, Shiga, Japan). The assays were performed on a LightCycler 96 quantitative PCR instrument (ABI QuantStudio 6 Pro, Thermo Fisher Scientific) and a 96-well optical plate [18].

Mouse brain sections were washed three times with PBS. They were blocked in PBS containing 10% normal goat serum (240528; Affinity Biosciences Ltd., Changzhou, Jiangsu, China) for 1 h at room temperature, then incubated overnight at 4 °C with the primary antibody. After washing with PBS, sections were incubated in blocking buffer with secondary antibodies for 1 h. Cell nuclei were stained with Hoechst 33342 (C0031-01, Solarbio) for 8 min at room temperature. Following three PBS washes, images were acquired using a confocal microscope (FV3000, Olympus, Tokyo, Japan). Primary antibodies used for immunostaining were MAP2 (rabbit; 1:200, Cell Signaling Technology) and NeuN (rabbit, 1:200, Cell Signaling Technology). Secondary antibodies were Alexa Fluor 488 streptavidin anti-biotin (1:400, 702410ES03, Yeasen Biotechnology, Shanghai, China) and Alexa Fluor 488/555 coupled goat anti-rabbit IgG (1:800, cr3358443-5, Abcam).

The Morris water maze (MWM) was used to evaluate spatial learning and memory in mice. Briefly, a circular pool (diameter 120 cm, height 50 cm) filled with water (22 $\pm$ 1 °C, water depth 30 cm) was rendered opaque with non-toxic white pigment. A camera system was mounted vertically approximately 3 m above the pool. The pool was divided into four quadrants: east, west, south, and north, with a platform (diameter 12 cm, height 29 cm) placed centrally in the north quadrant. Distinctive visual cues (different colors/shapes) were fixed around the pool to aid platform localization. The 7-day test included three phases: visible platform training, hidden platform training, and probe trial (no platform). On day 1 (visible platform training), the platform was exposed 1 cm above the water, and each mouse was released into the water from the center of each quadrant (north, east, west, south). Latency to find the platform was recorded. Mice failing to locate the platform within 60 s were guided to the platform and allowed to stay for 15 s. A submerged platform (1 cm below the water) was used on days 2–6 (hidden platform training). Each mouse was trained four times per day and released from different quadrants in varying order. Escape latency (time to find the platform) was recorded, with training ending at the 60th second, and a 15-s temporary stay on the platform for mice failing to locate it. On day 7 (probe trial), the platform was removed. Spatial exploration was assessed by releasing mice from the quadrant opposite the original platform location. Data were analyzed by two-way repeated measures ANOVA.

Spatial memory was assessed using a Y-maze with removable arm partition baffles, as previously described [19]. Mice were habituated to the experimental environment for 1 week before the formal experiment. After habituation, distinct colored and shaped stickers were affixed to the end of each arm. The formal experiments commenced with each mouse tested for 10 min per session. The Y-maze experiment consisted of two phases, spaced 1 h apart. During the first phase of training, the novel arm was blocked with a partition baffle, and mice were placed at the end of the starting arm. After 10 min of free exploration at the starting arm and other accessible arms, mice were returned to their home cages. The second phase of the test was conducted 1 h later, with the partition baffle removed from the novel arm, and each mouse repositioned at the end of the starting arm. The time spent, distance traveled, and number of arm entries were recorded during 10 min of free exploration across all three arms.

Draxin expression data were acquired from the AD-dedicated database AlzData (http://www.alzdata.org/) by integrating five publicly available gene expression datasets (GSE28146, GSE29378, GSE36980, GSE48350, GSE5281). Samples were selected based on three stringent criteria: human hippocampal tissue, clear grouping into AD patients or age-matched normal controls, and $\ge$90% data integrity. The final cohort comprised 68 AD and 52 normal controls, with all AD cases conforming to the National Institute on Aging and Alzheimer’s Association (NIA-AA) diagnostic criteria. Dataset-specific sample sizes were as follows: GSE28146 (AD n = 12, control n = 8), GSE29378 (AD n = 15, control n = 10), GSE36980 (AD n = 14, control n = 9), GSE48350 (AD n = 13, control n = 12), and GSE5281 (AD n = 14, control n = 13). Data analysis was performed using R software (v4.3.0, R Core Team, Vienna, Austria). Quantile normalization was applied to eliminate inter-dataset batch effects. The limma package (3.54.1, Gordon K. Smyth, Walter and Eliza Hall Institute of Medical Research, Melbourne, VIC, Australia) was used for linear model fitting to calculate draxin’s Log2 fold change (Log2FC) between the AD and control groups, two-tailed t-tests generated raw p-values, and the Benjamini-Hochberg method was employed to adjust for false discovery rate (FDR), with significance thresholds set at p 0.05 and FDR 0.2. To ensure the robustness and reliability of the integrated analysis, we performed additional reproducibility and sensitivity checks. Reproducibility check: we re-analyzed each dataset independently (rather than only integrating them) and confirmed consistent upregulation of draxin in AD hippocampi across all five datasets (p 0.05 for GSE28146, GSE29378, and GSE5281; p 0.1 for GSE36980 and GSE48350), supporting the robustness of the finding. Sensitivity analysis: exclusion of any single dataset did not alter the overall result (pooled p 0.05), indicating the integrated analysis was not biased by individual datasets.

Images were analyzed using ImageJ. All data were expressed as the mean $\pm$ SEM and analyzed using one-way ANOVA or two-way ANOVA for multiple comparisons. The significance level was set at p 0.05. The statistical software used in the study was GraphPad Prism 9.0 (GraphPad Software, Inc., San Diego, CA, USA) for Windows.

Following a search of the AD database AlzData—encompassing the publicly available datasets GSE28146, GSE29378, GSE36980, GSE48350, and GSE5281—draxin expression in the hippocampal region of AD patients was found to be higher than in normal subjects (p = 0.044; Fig. 1A). After considering data from multiple brain regions, no difference in the FDR was found (Fig. 1B). To investigate the role of draxin overexpression in the pathogenesis and progression of AD, we constructed a mouse model of draxin overexpression by injecting draxin-AAV into the bilateral hippocampus of WT mice. The spontaneous green fluorescent protein (GFP) green signal of the virus was visualized using fluorescence microscopy (Fig. 1C). A qRT-PCRassay showed that the level of draxin expression in the WT+AAV-draxin group was significantly increased relative to the WT+PBS and WT+AAV-NC groups (p 0.05; Fig. 1D).

Fig. 1.

Expression of draxin in the hippocampus of AD patients in AlzData. (A) Distribution of draxin expression in the hippocampus of AD patients and normal subjects (n = 68 in control group, n = 52 in AD group); (B) Draxin expression in the hippocampus of AD patients and normal subjects. (C) Fluorescence microscopy of the hippocampal gland of three groups of mice with different virus expression. Scale bar = 100 µm. (D) qRT-PCR detection of hippocampal draxin expression in three groups of mice. n = 4. * p 0.05; ns, p $>$ 0.05. (E) Fluorescence microscopy of draxin lentivirus after transfection of HT22 cells. Scale bar = 100 µm. (F) Draxin mRNA expression in cells from the three groups. n = 4. **** p 0.0001 for LV-draxin compared with the NC and LV-NC groups; ns for NC compared with the LV-NC control group, p $>$ 0.05. (G) Changes in cell viability after treatment of HT22 cells with different concentrations of A$\beta$42; **** p 0.0001; *** p 0.001; ** p 0.01. (H) Western blot detection of changes in the expression of p-Tau396 after treatment of HT22 cells with an A$\beta$42 concentration of 20 µM, n = 4; (I) Quantification of changes in the expression of p-Tau396. n = 3. mean $\pm$ SEM, * p 0.05. AD, Alzheimer’s disease; FDR, false discovery rate; WT, wild-type; AAV, adeno-associated virus; PBS, phosphate-buffered saline; NC, negative control; LV-NC, lentivirus-negative control; A$\beta$42, amyloid-$\beta$-42; p-Tau, phosphorylated tau protein.

An AD cell model with overexpression of draxin was also constructed with HT22 cells. After mapping the optimal Multiplicity of Infection (MOI), draxin lentivirus was successfully transfected into HT22 cells, with the spontaneous GFP green signal of the virus visible by fluorescence microscopy (Fig. 1E). qRT-PCR revealed that draxin expression was significantly higher in the lentivirus (LV)-draxin group compared with the negative control (NC) and lentivirus-negative control (LV-NC) groups (p 0.0001; Fig. 1F), indicating successful construction of the HT22 cell model with draxin overexpression. To select the optimal A$\beta$ concentration, HT22 cells were treated with A$\beta$42 (0–50 µM) for 24 h, and their viability was then quantified using a CCK8 assay. A dose-dependent reduction in viability (p 0.05 vs control) was observed starting at 20 µM (p 0.01; Fig. 1G). After the treatment of cells with A$\beta$42 at a concentration of 20 µM, western blotting showed significantly increased tau protein phosphorylation (p 0.05; Fig. 1H,I). Therefore, a 20 µM concentration was selected to treat HT22 cells and AD cell model constructs.

To investigate whether draxin has a detrimental impact on hippocampal neuronal cells, the viability of cells in the NC, LV-NC, and LV-draxin groups was evaluated by CCK8 assay, while the apoptosis rate was assessed by Hoechst staining. The survival rate of cells in the LV-draxin group was significantly decreased compared with the LV-NC group (p 0.001; Fig. 2A), while the apoptosis rate was significantly increased (p 0.01; Fig. 2B). Similar results were observed in the AD cell model constructed with HT22 cells and A$\beta$42, with the survival rate of cells in the LV-draxin+A$\beta$42 group being significantly decreased, while the apoptosis rate was significantly increased (p 0.05; Fig. 2C,D). The expression levels of MAP2 and NeuN are shown in Fig. 2E. The expression of MAP2 and NeuN was lower in the LV-draxin group compared with the LV-NC+A$\beta$42 group, reaching significance for MAP2 (p 0.05; Fig. 2F).

Fig. 2.

Effects of draxin on the activity, neuronal-associated proteins, and Tau proteins of HT22 cells and AD cell models. (A) Hoechst staining of the three groups of cells. Scale bar = 200 µm; (B) Cell survival rate, as determined by CCK8 assay. Hoechst staining of the three groups of cells was used to calculate the apoptosis rate. (C) Hoechst staining of three AD cell models. Scale bar = 200 µm; (D) Cell viability, as determined by CCK8 assay, and apoptosis rate as determined by Hoechst staining of the three AD cell models. (E) Western blot detection of MAP2 and NeuN expression, and quantification of the results in three groups of HT22 cells, MAP2 n = 3, NeuN n = 4. (F) Western blot detection of MAP2 and NeuN expression, and quantification of the results in three groups of AD cell models, MAP2 n = 3, NeuN n = 3. (G) Western blot detection of p-Tau231 and p-Tau396 expression and quantification of the results in three groups of HT22 cells, p-Tau231 n = 4, p-Tau396 n = 4; (H) Western blot detection of p-Tau231 and p-Tau396 expression and quantification of the results in three groups of AD cell models, p-Tau231 n = 4, p-Tau396 n = 4. mean $\pm$ SEM, *** p 0.001; ** p 0.01, * p 0.05, ns, p $>$ 0.05. A$\beta$, amyloid-$\beta$; CCK8, Cell Counting Kit-8; MAP2, microtubule-associated protein 2; NeuN, neuronal nuclear antigen.

To determine whether draxin aggravated AD pathology, we examined the expression levels of p-tau in three groups of cells: NC, LV-NC, and LV-draxin. A trend for increased levels of phosphorylated tau proteins p-tau231 and p-tau396 was observed in the LV-draxin group in contrast to the LV-NC group (Fig. 2G). The level of phosphorylated tau protein p-tau396 in the NC group was significantly higher than in the LV-NC group (p 0.01), while p-tau231 showed a non-significant increase (p $>$ 0.05). The expression level of tau proteins in NC+A$\beta$42, LV-NC+A$\beta$42, and LV-draxin+A$\beta$42 cells is shown in Fig. 2H. The expression of p-tau231 was significantly increased in the LV-draxin+A$\beta$42 group compared with the LV-NC+A$\beta$42 group (p 0.05), while the expression of p-tau396 showed a non-significant increase (p $>$ 0.05). In the LV-NC+A$\beta$42 group, the expression of p-tau231 was significantly increased (p 0.01), while p-tau396 showed a non-significant increase (p $>$ 0.05) compared with the NC+A$\beta$42 group. Subsequently, we also evaluated the expression level of Wnt/$\beta$-catenin/GSK-3$\beta$ pathway components (Fig. 3A). The expression of Wnt3a and $\beta$-catenin was significantly downregulated in the LV-draxin group compared with the LV-NC group (p 0.05), while GSK-3$\beta$ and LRP6 expression was upregulated (p 0.05). The LV-NC group showed no significant differences in the expression of Wnt3a, $\beta$-catenin, GSK-3$\beta$, and LRP6 compared with the NC group (all p $>$ 0.05). Collectively, these findings suggest that overexpression of draxin in HT22 cells promotes the phosphorylation of tau proteins through activation of the Wnt/$\beta$-catenin/GSK-3$\beta$ pathway, especially when stimulated by A$\beta$42.

Fig. 3.

The effect of draxin on the levels of Wnt/$\beta$-catenin/GSK-3$\beta$ pathway-related proteins in HT22 cells and normal mouse hippocampus, as well as the expression of normal hippocampal neuron-related proteins and Tau protein phosphorylation levels. (A) Western blot detection of Wnt3a, $\beta$-catenin, GSK-3$\beta$, and LRP6 expression in three groups of cells, and quantification of the results, Wnt3a n = 3, $\beta$-catenin n = 4, GSK-3$\beta$ n = 3, LRP6 n = 3. (B,D,F,H) Immunofluorescence and Western blot detection of hippocampal MAP2 expression in three groups of mice, and quantification of the results, n = 4. (B,C): scale bar = 400 µm) (D,E,G,H,I) Immunofluorescence and Western blot detection of hippocampal NeuN expression in three groups of mice, and quantification of the results, n = 4; (J) Changes in the expression of LRP6 in three groups of mice, and quantification of the results, n = 4; (K) Western blot detection of Wnt/$\beta$-catenin/GSK-3$\beta$ pathway-related protein expression in three groups of mice, and quantification of the results, n = 4. mean $\pm$ SEM, * p 0.05; ** p 0.01; ns, p $>$ 0.05. GSK-3$\beta$, glycogen synthase kinase-3$\beta$; LRP6, lipoprotein receptor-related protein 6.

The Y-maze test and MWM were used to evaluate the learning and memory ability in the three groups of mice: WT+PBS, WT+AAV-NC, and WT+AAV-draxin. The spontaneous alternation rate (alternation %) of mice in the WT+AAV-draxin group was significantly lower than in the WT+AAV-NC group (p 0.01; Fig. 4A,B). The spatial recognition experiments revealed no significant differences between the three groups in terms of the number of new arm entries (number of new arms) and the percentage of exploration time in the new arm (time in the new arm%) (p $>$ 0.05; Fig. 4B).

Fig. 4.

Behavioral experiments aimed at detecting the effects of draxin on the learning and memory ability of normal mice. (A) Motion trajectory diagrams of mice from the three groups exploring the new arm of the Y maze spatial recognition experiment. (B) Y maze spontaneous alternation rate, number of times exploring the new arm, and percentage of time exploring the new arm by the three groups of mice in the alternation experiment; n = 10. ** p 0.01; ns, p $>$ 0.05. (C) Typical movement trajectory diagrams of the spatial exploration phase of the three groups of mice; Red dots = Mice starting points; Blue dots = Mice end points; (D) Time required to find the platform in the localization and navigation experiments of the three groups of mice (evasion latency); (E) Number of times traversing the platform in the spatial exploration phase, percentage of swimming distance in the quadrant where the platform was located during the exploration phase, and percentage of time spent on the platform in the spatial exploration phase of the three groups of mice. Percentage of swimming distance in the quadrant where the platform was located, and the time spent in the exploration phase where the platform was located; n = 10. * p 0.05, ** p 0.01 compared to the WT+AAV-NC group; ns, p $>$ 0.05; ^#p 0.05, ^#⁢#p 0.01, ^#⁢#⁢#p 0.001 compared to the WT+PBS group; ns, p $>$ 0.05.

In the MWM experiments, the time to find the platform (platform latency) of the mice decreased progressively over time during navigation training in the WT+AAV-NC and WT+PBS groups, but not in the WT+AAV-draxin group (Fig. 4C,D). In the probe experiment on day 6, mice in the WT+AAV-draxin group exhibited a significantly decreased number of platform crossings compared with the other two groups (p 0.001), as well as decreased time and distance in the target quadrant (p 0.05; Fig. 4E). These results suggest that draxin overexpression in the mouse brain leads to a decline in spatial learning and memory ability.

Neuropathological changes in the mouse brain were examined by detecting the expression of MAP2, NeuN, and p-tau, as well as the Wnt/$\beta$-catenin/GSK-3$\beta$ pathway (Fig. 3). Immunofluorescence (IF) revealed significantly decreased expression of MAP2 and NeuN in the WT+AAV-draxin group compared with the WT+AAV-NC group (p 0.05; Fig. 3B–E). Western blot results showed that MAP2 expression in the hippocampus of mice from the WT+AAV-draxin group was significantly decreased compared with the WT+AAV-NC group (Fig. 3F,H). There was no significant difference in the expression levels of MAP2 and NeuN between the WT+AAV-NC and WT+PBS groups.

The level of p-tau396 in mice from the WT+AAV-draxin group was increased compared with the WT+AAV-NC group (p 0.01; Fig. 3G,I–K), while no significant difference was observed between the other two groups (Fig. 3G,I). GSK-3$\beta$ expression was also increased in the WT+AAV-draxin group (p 0.05), while $\beta$-catenin showed a non-significant decrease, and LRP6 a non-significant increase (p $>$ 0.05). These results suggest that draxin overexpression in the mouse hippocampus promotes pathological hyperphosphorylation of tau protein, as well as neuronal damage that appears to be mediated by the Wnt/$\beta$-catenin/GSK-3$\beta$ pathway.

To explore the mechanisms underlying the detrimental impacts of draxin overexpression on the brain and neurons, RNA sequencing analysis was performed on HT22 cells from the LV-NC and LV-draxin groups. Significant changes were found in the expression of 66 genes, of which 31 were up-regulated and 35 were down-regulated (Fig. 5A). Heatmap clustering analysis showed different expression patterns among the samples (Fig. 5B). Gene ontology (GO) and kyoto encyclopedia of genes and genomes (KEGG) analyses revealed that the top 20 differentially expressed genes were significantly enriched in pathways for regulation of ribonuclease activity, adhesion of symbiont to host, response to protozoan, cellular response to interferon-beta, defense response to symbiont, defense response to virus, and regulation of nuclease activity. Amongst these, binding, cell part, and cellular process-related genes accounted for the highest percentage of biological processes (BP), cellular components (CC), and molecular functions (MF) (Fig. 5C).

Fig. 5.

RNA sequencing analysis of HT22 cells with draxin overexpression. (A) Differential gene volcano plot; (B) Clustering heatmap of differentially expressed genes (red color indicates up-regulated gene expression; blue color indicates down-regulated gene expression); (C) Differentially expressed gene GO function annotation map; (D) Differentially expressed gene GO enrichment bubble map; (E) Differentially expressed gene GO enrichment string map; (F) Differentially expressed gene KEGG function annotation map; (G) Differentially expressed gene KEGG enrichment bubble map; (H) Relative expression of BMF, AKR1B1 and AVEN genes in the cells of the two groups. n = 5. * p 0.05; ** p 0.01. GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; BMF, BCL2-modifying factor; AKR1B1, Aldo-keto reductase family 1 member B1.

Further analysis showed that draxin overexpression had a significant effect on hippocampal neuron glucose metabolism and alcohol metabolism, as well as regulating the stress response of aldehydes and the microvillus function of Schwann cells (Fig. 5D,E). KEGG enrichment analysis revealed that the differentially expressed genes were involved in a variety of metabolic regulations, with C5-branched-chain dicarboxylic acid metabolism (rich factor close to 1) being particularly significant. These genes were also enriched for biological processes such as the NF-$\kappa$B signaling pathway and extracellular matrix (ECM)-receptor interactions, and were associated with a variety of human diseases (Fig. 5F). Pathway analysis further revealed the important roles of the differentially expressed genes in glucose metabolism, lipid metabolism, amino acid metabolism, signaling, and the immune system (Fig. 5G). Finally, expression of the pro-apoptotic gene BCL2-modifying factor (BMF) was significantly elevated in the LV-draxin group (p 0.05), whereas expression of the anti-apoptotic genes aldo-keto reductase family 1 member B1 (AKR1B1) (p 0.01) and apoptosis and caspase activation inhibitor (AVEN) (p 0.05) were significantly reduced (Fig. 5H), suggesting that draxin may affect cell fate by regulating apoptosis-related genes.