2.1 Study participants

Fasting blood samples were collected from 11 to 12 normal children and 13 children with ASD at Chongqing Maternal and Child Health Hospital. The children with ASD included 9 males and 4 females (mean age ± SD = 5.7 ± 2.9 years), diagnosed based on the Diagnostic and Statistical Manual of Mental Disorders, 5th edition.²⁸ The normal children included 7 males and 5 females (mean age ± SD = 5.8 ± 1.5 years). These healthy children demonstrated no clinical findings suggestive of neuropsychiatric disorders. None of the participants had a recent history of infection or fever.

The study protocol was approved by the Medical Research Ethics Committee of Chongqing Medical University (IACUC-CQMU-2023-0334). Written informed consent was obtained from all parents in accordance with the principles of the Declaration of Helsinki (2008).

2.2 Enzyme-linked immunosorbent assay (ELISA)

Blood samples were collected in vacutainer tubes, held at 4°C for 30 min, and then centrifuged at 3500 rpm for 5 min to obtain serum samples. Human Quantikine ELISA Kits (manufactured by Wuhan ColorfulGene Biological Technology, China) were employed to detect serum levels of Sortilin and Tumor Necrosis Factor Alpha (TNF-α) following the manufacturer's protocol. PGRN levels in human serum and TNF-α levels in rat hippocampal tissue were measured using ELISA kits (produced by Boster Biological Technology, Wuhan, China).

2.3 Animal model

Animal procedures were evaluated and approved by the Ethics Committee of Experimental Animals of Chongqing Medical University and followed the Guide for the Care and Use of Laboratory Animals (NIH Publication No.85-23, revised 1996). Timed-pregnant Sprague–Dawley rats (Experimental Animal Center in Chongqing Medical University, Chongqing, China) were randomly assigned on gestational day (GD) 12.5 to receive intraperitoneal injections of either sterile saline or 600 mg/kg Sodium Valproate (dissolved in sterile saline). Male offspring of these rats were selected as the experimental subjects. The male offspring from dams receiving sterile saline were assigned to the Sham group, while the male offspring from dams receiving VPA were randomly divided into the VPA + vehicle group, VPA + PGRN group, VPA + Si-SORT1 group, and VPA + NC group for subsequent experiments.

2.4 Stereotaxic surgery

On PND10, all rats were administered inhalation anesthesia with 2.5% isoflurane and secured onto a stereotaxic apparatus (RWD Life Technology, Shenzhen, China). A microlitre syringe was inserted into each side of hippocampus using the following coordinates: 2.3–2.5 mm posterior, 2.5–2.8 mm lateral, 3–3.5 mm deep (relative to bregma). Rats were treated with the different 5 μL drug injections: 0.9% sterile saline (VPA + vehicle group), 50 ng/mL recombinant PGRN (r-PGRN) (SinoBiological, Beijing, China) dissolved in 0.9% sterile saline (VPA + PGRN group), SORT1 knockdown lentivirus (VPA + Si-SORT1 group), and scramble lentivirus (non-targeting sequences) (VPA + NC group). Rats of the Sham group underwent the same surgical procedure but did not receive any drug injections. On PND 17, the VPA + PGRN group was injected with rPGRN once again, using the same dosage as previously. Meanwhile, those in the Sham group and the VPA + vehicle group who matched with the VPA + PGRN group also received the same treatment as before. Criteria for postoperative analgesic treatment were predefined and included signs such as aggression, piloerection, hunched posture, or rapid and shallow breathing, in accordance with national and international guidelines.²⁹ Post-surgery, each rat was monitored for indicators like body weight loss from baseline, wound healing, and behavioral changes, porphyrin staining (a sign of stress, illness, or poor environmental conditions for rats). We employed a comprehensive animal welfare scoring table, adapted from Hampshire and co-workers,³⁰ to ensure objective and systematic assessment. These evaluations combined various parameters to identify welfare impairments and ascertain if humane endpoints were needed. If signs indicated insufficient pain relief or if a combined score exceeded 0.4, euthanasia would be carried out. In this study, no animals reached this threshold.

2.5 Gene knock-down

Purified plasmid DNA based on the pGLV1(U6/GFP) lentiviral vector drives a short hairpin DNA (shDNA) sequence (GCCCTGGAATTATGGAGAACA), was precisely engineered to target and silence the SORT1 gene in rats. The correct construction and insertion of the shDNA sequence were rigorously confirmed through sequencing methods.

2.6 Behavioral analysis

2.7 Western blot

Hippocampal tissues were isolated on ice and mechanically lysed using ice-cold RIPA buffer (Beyotime, Shanghai, China) supplemented with 1% (μL/μL) PMSF (Beyotime, Shanghai, China) and 1% (μL/μL) phosphatase inhibitor (Beyotime, Shanghai, China). After centrifugation at 12,000 rpm for 15 min at 4°C, the supernatants were collected. The protein concentration was quantified using the BCA assay (Beyotime, Shanghai, China). After adjusted to equal concentration, it was mixed with 5× SDS-PAGE loading buffer and subjected to boiling for denaturation. Subsequently, 20 μg of protein per lane was separated on 8–12% SDS-PAGE gel and transferred onto nitrocellulose membranes. Membranes were blocked with nonfat milk in TBST (Tris-buffered saline containing Tween-20) for 2 h and then incubated with primary antibodies overnight at 4°C. Primary antibodies used included anti-β-actin (1:100,000, ABclonal, China), anti-Synaptophysin (1:320,000, ProteinTech, China), anti-PSD95 (1:8000, CST, USA), anti-TNF-α (1:800, ABclonal, China), anti-IL-1β (1:1000, ABclonal, China), anti-p-IKKβ (1:1000, CST, USA), anti-IKKβ (1:1000, CST, USA), anti-p-p65 (1:1000, Affinity, USA), anti-p65 (1:1000, Affinity, USA), and anti-sortilin (1:1000, Abcam, UK). After removing excess antibody solution, membranes were washed with TBST and incubated with either goat anti-mouse IgG-HRP (1:5000, Boster, China) or goat anti-rabbit IgG-HRP (1:5000, Boster, China) for 2 h at room temperature. Detection was performed using an ECL kit (Abcam, UK) according to the manufacturer's instructions after washing off unconjugated secondary antibodies with TBST. Densitometric analysis of bands was conducted using ImageJ software (version 1.53c).

2.8 Immunofluorescence in vivo

Rats were euthanized and perfused with PBS to remove residual blood. The brains were fixed in 4% paraformaldehyde (PFA) for 24 h and washed before dehydration with graded sucrose solutions (10%, 20%, and 30%). The brains were then sectioned into 10-μm-thick coronal slices using a frozen microtome (Leica, Germany). Sections were permeabilized with 0.5% Triton-X in PBS for 30 min, blocked with 5% BSA for 1 h, and probed with primary antibodies NF-κB-p65 (1:500, Affinity, USA) and Iba1 (1:1000, Abcam, UK) overnight at 4°C. After washing, the sections were incubated with secondary antibodies Alexa Fluor®555 and Alexa Fluor®488 at room temperature for 1 h. Finally, sections were stained with DAPI (Sigma, USA). The morphology and colocalization of the sections were captured using a confocal microscope (Leica, Germany) and analyzed by ImageJ (version 1.53c). Morphological analysis and evaluation of microglia were carried out following methods described in relevant literature.^{31, 32}

2.9 Golgi-Cox staining

Golgi-Cox staining is a specialized technique for visualizing neuronal dendrite and dendritic spine morphology. All procedures were performed according to the protocols of Golgi-Cox OptimStain™ PreKit (Hitobiotch, USA). Brains were harvested as before, pretreated with the Kit and sliced into 100-μm-thick sections and dehydrated with different concentrations of alcohol. After stained, they were cleared in Xylene and covered up with neutral balsam. Dendritic spines of hippocampal neurons were observed and imaged using an Eclipse E100 microscope (Nikon, Japanese). For brain tissue, we employed a continuous sectioning method and sampled at consistent thickness intervals. From each rat, we used 3 slices, and for each slice, we selected 5 secondary dendrites, each with a length of 20–50 μm, for spine analysis using ImageJ software (version 1.53c).

2.10 Cell culture and drug treatment

BV2 microglia were cultured in sterile high-glucose DMEM containing 100 mL/L fetal bovine serum, 100 IU/mL penicillin, and 100 μg/mL streptomycin at 37°C and 50 mL/L CO2. Two days after cell passage, cells were divided into four groups: CON, VPA, VPA + Si-SORT1, and VPA + NC. The VPA + Si-SORT1 group and VPA + NC group were transfected with Si-SORT1 knockdown lentivirus and scramble lentivirus (non-targeting sequences), respectively, for 3 days. Subsequently, all groups except the CON group were treated with 2.5 mmol/L VPA, while the CON group was treated with an equivalent volume of PBS.

2.11 Immunofluorescence in vitro

Following the treatment described above, the cells were washed with pre-cooled PBS and fixed in pre-cooled 4% PFA for 30 min. The cells were then washed three times and blocked in 5% BSA for 1 h. Subsequently, the cells were probed with primary antibodies anti-PGRN (1:100, CST, USA) and anti-LAMP1 (1:100, Abcam, UK) overnight at 4°C, followed by incubation with goat anti-mouse Alexa Fluor®555 (1:100, Abcam, UK) and goat anti-rabbit Alexa Fluor®488 (1:100, Beyotime Biotechnology, China) secondary antibodies for 1 h at room temperature. The cells were then stained with DAPI and sealed with nail polish. Laser scanning confocal microscopy and ImageJ software mentioned before were used to analyze the colocalization of anti-PGRN and anti-LAMP1.

2.12 Statistical analysis

Statistical analysis was conducted using GraphPad Prism software (version 9.0.0). Data are expressed as mean ± SEM. Prior to selecting the appropriate statistical test, the assumptions of normality and homogeneity of variance were assessed. All data underwent normal distribution testing prior to analysis. Specifically, when the sample size was less than 50, we employed the Shapiro–Wilk test to assess normality. For sample sizes equal to or greater than 50, the D'Agostino-Pearson test was used to evaluate conformity to a normal distribution. For pairwise comparisons, an unpaired t-test was employed if the data conformed to normal distribution and satisfied the homogeneity of variance test; otherwise, the Mann–Whitney U test was utilized. When variance was not homogeneous, Welch's correction was applied. In the case of multiple group comparisons, one-way analysis of variance (ANOVA) with Sidak post hoc test was performed if the data followed a normal distribution. If the data failed to meet the homogeneity of variance test or normal distribution assumption, the Kruskal–Wallis test was implemented. A p-value less than 0.05 was deemed statistically significant.

3.1 PGRN alleviated autism-like behaviors in VPA-induced ASD rats

We initially examined the PGRN expression in the serum of ASD children using ELISA, and the PGRN expression in the hippocampus of VPA-induced ASD rats using western blot analysis (Figure 1A). The results revealed a significant reduction in the PGRN expression in the serum of ASD children (t = 3.229, df = 21, p = 0.0034 for ASD vs. Normal) and in the hippocampus of the VPA group at postnatal day (PND) 35 (t = 2.816, df = 22, p = 0.01 for VPA vs. CON). ASD rats were treated with recombinant PGRN twice, and behavioral tests were used to observe the changes in autism-like behaviors. The timeline of the surgical and behavioral tests is illustrated in Figure 1B.

The Juvenile Social Play test (Figure 1C) showed a notable increase in both social time (F_2,50 = 16.11, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + PGRN vs. VPA + vehicle) and dig time (F_2,50 = 3.795, p = 0.03; post hoc, p = 0.03 for VPA + vehicle vs. Sham, p = 0.05 for VPA + PGRN vs. VPA + vehicle) with PGRN supplementation. The Self-Grooming test (Figure 1D) demonstrated a significant reduction in the grooming duration with PGRN supplementation (F_2,43 = 13.46, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + PGRN vs. VPA + vehicle).

In the Open-Field test (Figure 1E,F), the VPA + PGRN group exhibited an increase in the number of climbing events, standing events, and grid crossing (climb: Kruskal–Wallis statistic = 19.94, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + PGRN vs. VPA + vehicle; stand: Kruskal–Wallis statistic = 13.11, p = 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p = 0.05 for VPA + PGRN vs. VPA + vehicle; central grids: Kruskal-Wallis statistic = 11.75, p = 0.003; post hoc, p = 0.35 for VPA + vehicle vs. Sham, p = 0.002 for VPA + PGRN vs. VPA + vehicle; circumambient grids: F_2,43 = 12.21, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + PGRN vs. VPA + vehicle).

In the sociability test stage of the three-chamber social test (Figure 1G), the VPA + PGRN group demonstrated an increase in the duration spent in chamber A (F_2,40 = 9.193, p = 0.0005; post hoc, p = 0.0012 for VPA + vehicle vs. Sham, p = 0.0007 for VPA + PGRN vs. VPA + vehicle) and a decrease in the time spent in chamber B (F_2,40 = 12.51, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + PGRN vs. VPA + vehicle). Concurrently, they exhibited an increased duration of interaction with stranger A (F_2,40 = 11.33, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p = 0.007 for VPA + PGRN vs. VPA + vehicle) and a decreased interaction with the object (Kruskal–Wallis statistic = 11.77, p = 0.0028; post hoc, p = 0.0022 for VPA + vehicle vs. Sham, p = 0.0113 for VPA + PGRN vs. VPA + vehicle). In the social novelty preference stage (Figure 1H), the VPA + PGRN group exhibited reduced time spent in chamber A (F_2,40 = 33.27, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + PGRN vs. VPA + vehicle) and increased time spent in chamber B (F_2,40 = 27.84, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + PGRN vs. VPA + vehicle). Additionally, they showed a prolonged duration of interaction with stranger B (F_2,40 = 11.29, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p = 0.01 for VPA + PGRN vs. VPA + vehicle).

These results suggest that PGRN alleviates social preference deficits, improves repetitive and rigid behavior, and enhances social competence in VPA-induced ASD rats.

To determine the extent of surgical damage, we monitored the body weight, general appearance, porphyrin staining, gait and posture, appetite, and wound condition of the animals pre-surgery and on postoperative days 1, 3, and 7. We found that the surgery did not cause significant trauma in the sham group, VPA + vehicle group, and VPA + PGRN group animals (see Table S1). Therefore, we believe that the behavioral changes observed are mainly influenced by the injection of PGRN, rather than by the surgery itself.

3.2 PGRN inhibits the NF-κB pathway and microglial activation in the hippocampus of VPA-induced ASD rats

We used enzyme-linked immunosorbent assay (ELISA) to detect TNF-α levels in the serum of normal and ASD children. The results revealed significantly elevated serum TNF-α levels in ASD children (t = 3.534, df = 22, p = 0.002 for ASD vs. Normal), indicating an activation of the inflammatory response in ASD. At the same time, we also found higher expression levels of TNF-α in the hippocampal tissue of the VPA + vehicle group compared to the Sham group, and PGRN could inhibit the increase of TNF-α in the hippocampal tissue (F_2,10 = 4.329, p = 0.0442; post hoc, p = 0.0499 for VPA + vehicle vs. Sham, p = 0.0489 for VPA + PGRN vs. VPA + vehicle) (Figure 2A). Subsequently, we used western blot analysis to assess the phosphorylation levels of IKKβ and p65 in the hippocampus of different groups of ASD rats at PND35 (Figure 2B–D). The results showed that PGRN supplementation significantly mitigated the increased levels of IKKβ and p65 phosphorylation in the hippocampus (IKKβ: F_2,57 = 5.121, p = 0.009; post hoc, p = 0.02 for VPA + vehicle vs. Sham, p = 0.01 for VPA + PGRN vs. VPA + vehicle; p65: F_2,30 = 4.172, p = 0.03; post hoc, p = 0.04 for VPA + vehicle vs. Sham, p = 0.04 for VPA + PGRN vs. VPA + vehicle). Moreover, the expression levels of the pro-inflammatory cytokines TNF-α and IL-1β in the VPA + PGRN group exhibited a notable reduction compared to the VPA + vehicle group (F_2,24 = 11.45, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + PGRN vs. VPA + vehicle; F_2,45 = 6.219, p = 0.004; post hoc, p = 0.003 for VPA + vehicle vs. Sham, p = 0.03 for VPA + PGRN vs. VPA + vehicle). Furthermore, we performed immunofluorescence staining to observe the subcellular localization of the p65 protein in the hippocampus of the different groups at PND35 (Figure 2E), which revealed reduced nuclear localization of the p65 protein in the VPA + PGRN group compared to the VPA + vehicle group. These results indicate that PGRN can reduce inflammation in the hippocampus.

To further explore whether PGRN regulates microglial activation, we performed immunofluorescence staining to examine the morphology and quantity of microglia in the hippocampus of the three groups of rats at PND35. ImageJ software was used to analyze the morphology of Iba-1 positive cells in each group, including binary processing and noise reduction, outline extraction, skeleton analysis, and fractal dimension analysis, the results of which were labeled as binary, outline, skeleton, and hull and circle, respectively (Figure 3A). We applied principal component analysis (PCA) to evaluate eight indicators related to microglial morphology (Figure 3B). The analysis extracted the first three principal components: PC1, PC2, and PC3. These accounted for 38.58%, 20.93%, and 18.31% of the total variance, respectively, cumulatively contributing to 77.82% of the variance. Two biplots were utilized to provide a comprehensive display of both the loadings of the eight indicators and the PC scores of microglia from each group. The results distinctly indicated that microglia in the VPA + vehicle group exhibited a rounder cell body, shorter branches, and a more complex typing dimension. This suggests that the cells in this group were more akin to an amoeboid activation state. In contrast, microglial activation appeared to be inhibited following PGRN supplementation.

We also assessed the microglial cell counts in the hippocampus of the three groups (Figure 3C). The Iba1-positive cell count was remarkably increased in the VPA + vehicle group compared to the Sham group, while it was significantly decreased in the VPA + PGRN group (F_2,64 = 58.8, p < 0.001; post hoc, p < 0.001 for VPA vs. CON, p < 0.001 for VPA + PGRN vs. VPA + vehicle, Figure 3K). These findings provide additional evidence that the activated microglia can be inhibited by PGRN supplementation.

Based on the aforementioned results, PGRN can inhibit microglial activation and the NF-κB pathway in the hippocampus of VPA-induced ASD rats.

3.3 PGRN alleviated dendritic spine injury in the hippocampus of VPA-induced ASD rats

To explore whether PGRN can influence the synaptic development that occurred following the inhibition of the NF-kB pathway and microglial activation, we employed Golgi-Cox staining to examine the density and morphology of secondary dendritic spines of the hippocampal neurons in the three groups of rats at PND35 (Figure 4A). The total density of dendritic spines and the density of mushroom-shaped dendritic spines in the hippocampal neurons exhibited a significant decrease in the VPA + vehicle group, however, both these parameters were significantly increased in the VPA + PGRN group (total dendritic spines: F_2,109 = 9.393, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p = 0.006 for VPA + PGRN vs. VPA + vehicle; mushroom-shaped dendritic spines: F_2,109 = 7.854, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p = 0.01 for VPA + PGRN vs. VPA + vehicle, Figure 4B). Furthermore, we conducted western blotting to assess the expression of synaptic development-related proteins, PSD95 and SYP, in the hippocampus (Figure 4C). In the VPA + vehicle group, the decreased levels of PSD95 and SYP partially rebounded following PGRN supplementation (PSD95: F_2,57 = 8.887, p < 0.001; post hoc, p = 0.002 for VPA + vehicle vs. Sham, p < 0.001 for VPA + PGRN vs. VPA + vehicle; SYP: Kruskal-Wallis statistic = 12.9, p = 0.002; post hoc, p = 0.01 for VPA + vehicle vs. Sham, p = 0.002 for VPA + PGRN vs. VPA + vehicle, Figure 4D). This observation was in line with the findings from the Golgi–Cox staining, further substantiating that PGRN supplementation effectively alleviated the dendritic spine injury in VPA-induced ASD rats.

3.4 Suppression of the elevated sortilin level upregulated the PGRN expression in the hippocampus of VPA-induced ASD rats

In the present study, ELISA results showed higher serum sortilin levels in ASD children compared to normal children (t = 3.534, df = 22, p = 0.0019 for normal vs. ASD, Figure 5A). Furthermore, we found a significant increase in the sortilin expression level in the hippocampus of the VPA group at PND7, PND14, and PND35 (PND7:t = 2.455, df = 20, p = 0.0234 for VPA vs. CON; PND14: t = 2.172, df = 16, p = 0.0452 for VPA vs. CON; PND35:t = 3.192, df = 18, p = 0.0051 for VPA vs. CON, Figure 5C).

Next, inhibition of the sortilin expression led to a significant reduction in the PGRN expression in the VPA + vehicle group and significantly increased the PGRN expression in the VPA + Si-SORT1 group (F_3,96 = 9.825, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p = 0.04 for VPA + Si-SORT1 vs. VPA + vehicle; p = 0.88 for VPA + NC vs. VPA + vehicle, Figure 5E).

To accurately determine the impact of sortilin on PGRN, we cultured the BV2 microglia cells in vitro and then treated them with VPA (2.5 mmol/L) to simulate the effects of VPA on the microglia in the nervous system. Immunofluorescence staining was performed to visualize the localization of anti-PGRN and anti-LAMP1 (Figure 5F), followed by co-localization analysis (Figure 5G–I). The VPA + Si-SORT1 group displayed a significant reduction in the Manders' M1, Manders' M2, and Pearson's R values (Pearson's R value: Kruskal-Wallis statistic = 96.56, p < 0.001; post hoc, p < 0.001 for VPA vs. CON, p < 0.001 for VPA + Si-SORT1 vs. VPA; p > 0.99 for VPA + NC vs. VPA; Manders' M1: Kruskal–Wallis statistic = 61.28, p < 0.001; post hoc, p < 0.001 for VPA vs. CON, p < 0.001 for VPA + Si-SORT1 vs. VPA; p = 0.25 for VPA + NC vs. VPA; Manders' M2: Kruskal-Wallis statistic = 42.55, p < 0.001; post hoc, p < 0.001 for VPA vs. CON, p < 0.001 for VPA + Si-SORT1 vs. VPA; p > 0.99 for VPA + NC vs. VPA). These findings suggest that inhibition of the sortilin expression reduced the PGRN degradation by lysosomes, leading to an increase in the PGRN protein levels.

3.5 Suppression of sortilin expression inhibited the microglial activation and inflammation in the hippocampus of VPA-induced ASD rats

To determine whether sortilin inhibition promoted microglial activation and inflammation, we continued to use immunofluorescence staining to conduct the morphological analysis of microglia in the hippocampus (Figure 6A), and applied PCA to evaluate microglial morphology (Figure 6B). The analysis extracted the first three principal components: PC1, PC2, and PC3, which accounted for 43.81%, 22.38%, and 13.55% of the total variance, respectively, cumulatively contributing to 79.74% of the variance. The biplots distinctly indicated that after SORT1 knockdown, microglial activation appeared to be inhibited.

Moreover, the western blot analysis demonstrated a notable decrease in both IL-1β and TNF-α levels following SORT1 knockdown (Figure 6C–E) (IL-1β: Kruskal-Wallis statistic = 16.66, p < 0.001; post hoc, p = 0.004 for VPA + vehicle vs. Sham, p = 0.01 for VPA + Si-SORT1 vs. VPA + vehicle; p > 0.99 for VPA + NC vs. VPA + vehicle; TNF-α: F_3,64 = 2.233, p = 0.09; post hoc, p = 0.009 for VPA + vehicle vs. Sham, p = 0.05 for VPA + Si-SORT1 vs. VPA + vehicle; p = 0.46 for VPA + NC vs. VPA + vehicle).

These results indicate that sortilin inhibition can inhibit microglial activation and reduce inflammation in VPA-induced ASD rats.

3.6 Suppression of sortilin expression alleviated autism-like behavior in VPA-induced ASD rats

To examine whether sortilin suppression alleviates autism-like behaviors, we used the same behavioral testing protocol as previously described to assess the autism-like behaviors in the four groups of rats during PND28-32. The timeline of this evaluation is depicted in the corresponding Figure 7A.

The juvenile social play test (Figure 7B) showed a notable increase in social time with SORT1 knockdown (F_3,44 = 8.728, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p = 0.002 for VPA + Si-SORT1 vs. VPA + vehicle, p = 0.74 for VPA + NC vs. VPA + vehicle). The Self-Grooming test (Figure 7C) demonstrated a significant reduction in the grooming duration with SORT1 knockdown (Kruskal-Wallis statistic = 34.86, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + Si-SORT1 vs. VPA + vehicle, p > 0.99 for VPA + NC vs. VPA + vehicle). Furthermore, in the open-field test (Figure 7D,E), the VPA + Si-SORT1 group exhibited an increase in the number of climbing events, circumambient grid crossings, and central grid crossings (climb: Kruskal-Wallis statistic = 25.93, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p = 0.008 for VPA + Si-SORT1 vs. VPA + vehicle, p > 0.99 for VPA + NC vs. VPA + vehicle; circumambient grids: F_3,45 = 33.01, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + Si-SORT1 vs. VPA + vehicle, p = 0.96 for VPA + NC vs. VPA + vehicle; central grids: Kruskal-Wallis statistic = 28.1, p < 0.001; post hoc, p = 0.02 for VPA + vehicle vs.Sham, p = 0.04 for VPA + Si-SORT1 vs. VPA + vehicle, p = 0.18 for VPA + NC vs. VPA + vehicle).

During the first stage of the three-chamber social test, the VPA + Si-SORT1 group exhibited a marked increase in the time spent in chamber A and interaction with stranger A, while the time spent in chamber B was significantly decreased (chamber A: F_3,36 = 10.25, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + Si-SORT1 vs. VPA + vehicle, p < 0.001 for VPA + NC vs. VPA + vehicle; stranger A: F_3,36 = 9.724, p < 0.001; post hoc, p = 0.002 for VPA + vehicle vs. Sham, p < 0.001 for VPA + Si-SORT1 vs. VPA + vehicle, p = 0.08 for VPA + NC vs. VPA + vehicle; chamber B: F_3,36 = 12.38, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + Si-SORT1 vs. VPA + vehicle, p < 0.001 for VPA + NC vs. VPA + vehicle, Figure 7F). During the second stage, the time spent in chamber A was notably decreased, while the time spent in chamber B and the duration of interaction with stranger B were significantly increased (chamber A: F_3,36 = 5.296, p = 0.004; post hoc, p = 0.009 for VPA + vehicle vs. Sham, p = 0.003 for VPA + Si-SORT1 vs. VPA + vehicle, p = 0.2 for VPA + NC vs. VPA + vehicle; chamber B: F_3,36 = 5.752, p = 0.003; post hoc, p = 0.005 for VPA + vehicle vs. Sham, p = 0.002 for VPA + Si-SORT1 vs. VPA + vehicle, p = 0.13 for VPA + NC vs. VPA + vehicle; stranger B: Kruskal-Wallis statistic = 28.5, p < 0.001; post hoc, p < 0.001 for VPA + vehicle vs. Sham, p < 0.001 for VPA + Si-SORT1 vs. VPA + vehicle, p > 0.99 for VPA + NC vs. VPA + vehicle, Figure 7G).

Collectively, these results suggest that sortilin inhibition can reduce repetitive and rigid behavior, enhance sociability, and mitigate impaired social preference in VPA-induced ASD rats.

We also monitored the body weight, general appearance, porphyrin staining, gait and posture, appetite, and wound condition of the animals pre-surgery and on postoperative days 1, 3, and 7 to exclude any traumatic damage to rats during the Lentivirus injection process. We found that the surgery did not cause significant trauma in the sham group, VPA + vehicle group, VPA + si-SORT1 group, and VPA + NC group animals (see Table S2). Therefore, we believe that the behavioral changes observed are mainly influenced by the knockdown of SORT1, rather than by the surgery itself.