Altered Gut Microbial Diversity and Depletion of SCFA-Producing Taxa Associated with ASD-like Phenotypes in a Prenatal VPA Rat Model

(1) Causal Mechanism Ambiguity: Most studies establish correlations between gut dysbiosis and ASD symptoms but lack mechanistic clarity. For instance, while reduced SCFA levels are observed in VPA models, whether this is a primary driver of neuroinflammation or a secondary effect of microbial dysbiosis remains unaddressed [10]. (2) Model Translational Limitations: Many rodent studies focus on acute VPA exposure, whereas human ASD risk involves prolonged prenatal exposure. The chronic effects of VPA on microbial-SCFA-neuroinflammation cascades across developmental stages are understudied [11]. (3) Therapeutic Target Gaps: Current interventions targeting the microbiota (e.g., probiotics, prebiotics) lack specificity due to an incomplete understanding of microbial-taxa-SCFA interactions. Identifying key SCFA-producing taxa and their neuroactive metabolites is critical for developing precision therapies [12].

(1) Mechanistic Dissection: Using 16S rRNA sequencing, we systematically characterize the gut microbiota in VPA-exposed rats, distinguishing causal relationships from correlations. (2) Translational Model Refinement: By employing a clinically relevant VPA dosing regimen (400–450–400 mg/kg on GD11.5–12.5–13.5), we mimic human first-trimester exposure, enhancing model relevance. (3) Multi-Omics Integration: Combining behavioral assays (open field, three-chamber social interaction), neuroinflammation profiling (ELISA, immunofluorescence), and oxidative stress markers, we establish a comprehensive framework linking microbial-metabolite-neuronal pathways.

Prenatal VPA exposure disrupts gut microbial diversity, particularly reducing SCFA-producing taxa, which leads to systemic and central neuroinflammation via impaired histone deacetylase regulation and oxidative stress, ultimately driving ASD-like behavioral phenotypes.

This study, by identifying SCFA-producing taxa as critical mediators, could unlock novel therapeutic strategies, such as targeted microbial supplementation or histone deacetylase modulation, which are more precise than current non-specific interventions. This work not only advances our understanding of ASD etiology but also provides a translational framework for developing microbiota-based therapies.

The open field test was employed to evaluate spontaneous exploratory behaviors and anxiety-related responses in experimental animals within a novel environment.

As shown in Figure 1, valproic acid (VPA) exposure significantly suppressed exploratory behaviors while enhancing anxiety-like responses in Sprague–Dawley (SD) rats. The specific findings are detailed below:

The marble-burying test was employed to evaluate repetitive stereotypic behaviors in rodents. Valproic acid (VPA)-exposed Sprague–Dawley (SD) rats exhibited significantly elevated marble-burying counts compared to controls (Figure 1E, p < 0.01). Representative images of marble-burying outcomes are shown in Figure 1G.

VPA-exposed rats demonstrated significantly increased self-grooming frequency relative to controls (Figure 1D, p < 0.05).

Conclusions:

VPA-exposed SD rats displayed enhanced repetitive stereotypic behaviors, including increased marble-burying and self-grooming activities. These findings suggest neurodevelopmental anomalies associated with basal ganglia-cortical circuit dysfunction.

The three-chamber test evaluated social motivation (0–10 min) and social novelty preference (10–20 min) through timed exploration measurements between unfamiliar conspecifics (Stranger 1/2) and non-social stimuli (empty cages/objects).

VPA-exposed SD rats exhibited impaired social motivation compared to controls (Figure 2A).

The VPA group spent significantly less time exploring the Stranger 1 cage (p < 0.01) while displaying increased preference for the empty cage (p < 0.01).

Reduced sniffing durations were observed in VPA-exposed rats for both Stranger 1 (p < 0.001) and objects (p < 0.001). Representative movement trajectories are shown in Figure 2C (ASD model) and Figure 2D (control).

Conclusions: VPA-exposed rats demonstrated avoidance of social stimuli (Stranger 1) and preferential exploration of non-social contexts, indicating gestational VPA-induced social competence deficits (p < 0.01).

VPA-exposed rats exhibited prolonged interaction with the familiar Stranger 1 cage (p < 0.001) but reduced exploration of the novel Stranger 2 cage (p < 0.001).

Increased sniffing time toward Stranger 1 (p < 0.0001) contrasted with diminished investigation of Stranger 2 (p < 0.01). Representative trajectories are displayed in Figure 2E (ASD model) and Figure 2F (control).

Conclusions: VPA-exposed SD rats displayed restricted social interest and impaired novelty preference, suggesting compromised cognitive flexibility.

The Morris water maze (MWM) task evaluates spatial learning and memory through acquisition trials and probe trials.

To characterize the neuroinflammatory response in the prefrontal cortex following prenatal VPA exposure, we quantified cytokine profiles using enzyme-linked immunosorbent assays (ELISA). As shown in Figure 4, VPA-exposed Sprague–Dawley rats exhibited a pronounced proinflammatory shift, with significantly elevated levels of interleukin-1β (p < 0.0001, Figure 4A), substantially increased concentrations of interleukin-6 (p < 0.0001, Figure 4B), and markedly augmented tumor necrosis factor-α expression (p < 0.001, Figure 4C) compared to saline-treated controls. Concurrently, we observed significant suppression of the anti-inflammatory cytokine interleukin-10 in VPA-exposed animals (p < 0.0001, Figure 4D), indicating a comprehensive disruption of neuroimmune homeostasis. These findings collectively demonstrate that prenatal VPA exposure induces a neuroinflammatory imbalance in the prefrontal cortex through coordinated upregulation of proinflammatory mediators and downregulation of anti-inflammatory signaling pathways.

To evaluate oxidative stress responses in the prefrontal cortex following prenatal VPA exposure, we quantified key antioxidant enzymes and oxidative damage markers. As shown in Figure 4, VPA-exposed rats exhibited significant suppression of antioxidant defenses, with markedly diminished glutathione peroxidase activity (p < 0.0001, Figure 4E), substantially reduced glutathione concentrations (p < 0.001, Figure 4F), significantly lower superoxide dismutase levels (p < 0.0001, Figure 4G), and profoundly decreased catalase activity (p < 0.0001, Figure 4K). Concurrently, we observed pronounced elevation of oxidative damage indicators, including significantly increased malondialdehyde accumulation (p < 0.001, Figure 4H), markedly augmented total nitric oxide synthase activity (p < 0.0001, Figure 4I), and substantially elevated nitric oxide concentrations (p < 0.01, Figure 4J). These results demonstrate that prenatal VPA exposure induces systemic redox imbalance characterized by compromised antioxidant capacity and exacerbated oxidative damage in the prefrontal cortex.

To assess the pathological features of neuroinflammation in the group of SD rats exposed to VPA, the expression level of the microglial marker Iba1 was detected via immunofluorescence staining in this study. As shown in Figure 5A–G, the fluorescence intensity of Iba1 in the hippocampal CA1 region of SD rats was much lower in the control group than in the group of SD rats exposed to VPA (p < 0.0001). Similarly, as shown in Figure 5H–N, the fluorescence intensity of Iba1 in the prefrontal cortex of SD rats was much lower in the control group than in the group of SD rats exposed to VPA (p < 0.0001). The results revealed that microglial Iba1 expression was significantly elevated in the prefrontal cortex and hippocampal CA1 region of SD rats in the group of SD rats exposed to VPA compared with those in the normal control group.

The above results indicated that the group of SD rats exposed to VPA successfully induced the transformation of microglia from a resting state to a proinflammatory phenotype, suggesting that neuroinflammation may be involved in the pathological process of ASD-related behavioral abnormalities.

To characterize gut microbiota alterations in the VPA-induced ASD model, we conducted hierarchical taxonomic analysis across multiple levels. As illustrated in Figure 6A, class-level profiling revealed a significant increase in Bacteroidia within the ASD model group compared to controls, suggesting potential dysregulation in carbohydrate metabolism pathways. At the family level (Figure 6B), we observed depletion of SCFA-producing taxa, including Prevotellaceae, Ruminococcaceae, and Lachnospiraceae. The reduction in Prevotellaceae, which is functionally associated with mucin degradation, implies compromised gut mucosal integrity. Furthermore, genus-level composition (Figure 6C) demonstrated diminished abundances of Prevotella and Ruminococcus in ASD models, concomitant with increased Lactobacillus and Escherichia_Shigella. The enrichment of Escherichia_Shigella, a genus linked to pro-inflammatory activity, aligns with the observed inflammatory phenotype. Order-level analysis (Figure 6D) confirmed significant suppression of Bifidobacteriales in ASD animals. At the phylum level (Figure 6E), we documented a marked decrease in Bacteroidota (Bacteroidetes) abundance.

Microbial diversity assessment revealed substantial structural divergence between groups. Venn diagram analysis (Figure 6F) identified 275 shared operational taxonomic units (OTUs), with 586 ASVs unique to controls and 284 unique to ASD models. This reduced taxonomic overlap indicates a distinct microbial community architecture in ASD rats, characterized by depletion of native taxa and emergence of ASD-associated species. Collectively, these hierarchical shifts depict an ASD-associated gut microbiota profile featuring depletion of SCFA-producing taxa (particularly Firmicutes) and reduced microbial diversity, supporting gut dysbiosis as a pathophysiological hallmark that may contribute to the metabolic and immune dysfunction observed in neurodevelopmental disorders.

At the family level (): Key taxonomic shifts included elevated Enterobacteriaceae and Erysipelotrichaceae in ASD rats, alongside depletion of Lachnospiraceae and Christensenellaceae. Figure S1A

At the genus level (): Lactobacillus, and Roseburia were reduced in the ASD group. Figure S1B

At the order level (): The ASD group exhibited pronounced taxonomic imbalances. Enrichment of Enterobacterales, Erysipelotrichales, and Staphylococcales contrasted with reduced Lachnospirales, Bifidobacteriales. These opposing trends may reflect disrupted microbial functional networks critical for gut homeostasis. Figure S1C

At the phylum level (): Hierarchical clustering revealed divergent bacterial community structures between the control group and the VPA-induced ASD model rats. Specific bacterial phyla displayed altered relative abundances in the ASD group, as indicated by heatmap color gradients, suggesting a broad reorganization of the gut microbiota. Pseudomonadota and Thermodesulfobacteriota were elevated in the ASD group. Figure S1D

To elucidate taxonomic shifts associated with ASD-like phenotypes, we performed multi-level phylogenetic analyses. The family-level cladogram () revealed significant enrichment of Peptostreptococcales-Tissierellales (order) and Peptostreptococcaceae (family)—taxa implicated in gut dysbiosis and inflammation—in the VPA-exposed group. Concurrently, Enterobacteriaceae (family) within class Gammaproteobacteria exhibited pronounced elevation, consistent with established patterns of proteobacterial overgrowth in neurodevelopmental disorders. Linear discriminant analysis at the family level () further identified Gammaproteobacteria (class) and Enterobacterales (order) as dominant discriminators in ASD models, whereas Christensenellaceae and Anaerovoracaceae were enriched in controls. This microbial shift suggests a transition toward pro-inflammatory or opportunistic communities in ASD pathogenesis. Figure S2A Figure S2B

Genus-level characterization () demonstrated distinct clustering of Enterobacterales and Desulfovibrionales in ASD models, contrasting with Lachnospiraceae predominance in controls. Supporting this, LDA scoring () confirmed significant enrichment of Escherichia_Shigella and c_Gammaproteobacteria in the ASD group, with the prominence of Escherichia_Shigella aligning mechanistically with dysbiosis-driven inflammatory pathways. At the phylum level, cladogram analysis () revealed broad structural reorganization marked by significant enrichment of Pseudomonadota—a phylum linked to mucosal colonization and immunomodulation—indicating systemic microbiota restructuring. This finding was corroborated by phylum-level LDA scoring (), where Pseudomonadota exhibited the highest discriminative power between groups, solidifying its role as a biomarker for ASD-associated dysbiosis. Figure S2C Figure S2D Figure S2E Figure S2F

Comparative analysis of differentially abundant microbial taxa at the family (), genus (), order (), and phylum () levels. Figure S3A Figure S3B Figure S3C Figure S3D

Family level (): Compared to the control group, VPA-induced ASD model rats exhibited an increased relative abundance of Enterobacteriaceae, while Lachnospiraceae and Ruminococcaceae were significantly reduced. The enrichment of Enterobacteriaceae suggests a potential association with intestinal barrier dysfunction and inflammatory responses. Conversely, the depletion of Lachnospiraceae and Ruminococcaceae—taxa linked to short-chain fatty acid (SCFA) production—may reflect impaired microbial diversity and compromised metabolic homeostasis. Figure S3A

Genus level (): The VPA group showed elevated levels of Escherichia_Shigella, accompanied by reduced abundances of Bifidobacterium. The rise in Escherichia_Shigella implies dysbiosis-associated pathogenicity. The decline in Bifidobacterium aligns with diminished microbial functionality in maintaining gut homeostasis. Figure S3B

Order level (): A marked increase in Enterobacterales and a reduction in Bifidobacteriales were observed in the VPA group. The dominance of Enterobacterales reinforces the trend of pro-inflammatory microbial shifts, while the depletion of Bifidobacteriales highlights a potential loss of beneficial taxa critical for gut immune regulation. Figure S3C

Phylum level (): VPA-treated rats displayed an elevated abundance of Pseudomonadota. The expansion of Pseudomonadota, a phylum containing facultative anaerobes, may indicate environmental stress adaptation in the gut ecosystem. Figure S3D

Shannon Rarefaction Curve (): The rarefaction curve for the control group (blue) plateaued at higher sequencing depths, indicating robust species richness and evenness. In contrast, the VPA-induced ASD model group (red) reached saturation earlier at lower sequencing depths, with an overall shallower curve, suggesting diminished microbial diversity. This trend implies a less complex and more homogenized gut microbiota structure in the ASD model. Figure S4A

Alpha Diversity Metrics (): Consistently, the VPA-induced ASD model group exhibited lower median values across all alpha diversity indices compared to controls (). Specifically, the Chao1 index () and Observed Features (), which estimate species richness, displayed narrower distributions and reduced medians in the ASD model group, indicative of decreased microbial richness. Similarly, the Faith PD index (), reflecting phylogenetic diversity, showed a decline in evolutionary breadth among microbial taxa. The Shannon entropy () and Simpson index (), both integrating richness and evenness, further highlighted reduced community evenness and dominance of fewer taxa in the ASD model group. Collectively, these results suggest a significant contraction in microbial diversity and ecological complexity following VPA induction. Figure S4B–F Figures S4B–F Figure S4B Figure S4D Figure S4C Figure S4E Figure S4F

Bray–Curtis Dissimilarity (): The Bray–Curtis dissimilarity index, which quantifies compositional differences based on species abundance, exhibited distinct patterns between groups (). The control group displayed a narrower distribution and lower median dissimilarity values (), whereas the VPA-induced ASD model group showed a broader distribution and elevated median dissimilarity (). This suggests heightened variability in microbial community structure among ASD model individuals compared to controls. Figure S5A,B Figure S5A,B Figure S5A Figure S5B

Unweighted UniFrac Distance (): Unweighted UniFrac analysis, which accounts for phylogenetic presence/absence differences, further corroborated these trends. The control group exhibited lower median distances and tighter clustering (), while the ASD model group displayed significantly higher median distances and greater dispersion (). This indicates that microbial lineages in the ASD model group diverged phylogenetically from the control group, independent of abundance variations. Figure S5C,D Figure S5C Figure S5D

Weighted UniFrac Distance (): Weighted UniFrac analysis, incorporating both phylogenetic relationships and species abundance, revealed similar patterns. The control group maintained lower median distances (), whereas the ASD model group showed elevated distances with increased variability (). These results highlight that ASD-associated microbial shifts encompass both phylogenetic and abundance-level alterations, underscoring a multifaceted dysbiosis. Figure S5E,F Figure S5E Figure S5F

The network correlation analysis () revealed a complex interaction pattern among various bacterial families within the microbial community (). Notably, Actinomycetota and Bacillota exhibited numerous interactions, suggesting a high level of co-occurrence and potential synergistic relationships (). Additionally, Bacteroidota and Pseudomonadota also showed significant interactions, indicating their possible functional interdependencies (). Figure S6A–D Figure S6A Figure S6B Figure S6C,D

The observed interaction patterns provide valuable insights into the structure and function of the microbial community. The frequent interactions between Actinomycetota and Bacillota may reflect their complementary roles in nutrient cycling and environmental adaptation. Similarly, the interactions between Bacteroidota and Pseudomonadota could be indicative of their collaborative efforts in degrading complex organic matter. Further research is warranted to elucidate the specific mechanisms underlying these interactions and their implications for ecosystem health and stability.

All procedures were conducted in accordance with the Institutional Animal Care Committee guidelines of Guangzhou Institute of Biomedicine and Health (Approval No. IACUC-2022-0063) under AAALAC-accredited SPF conditions.

SD rats were placed in the experimental room one hour prior to the procedures to acclimate to the laboratory conditions.

We assessed sociability and social novelty preference using a rectangular polycarbonate arena (120 × 40 × 40 cm) divided into three equal compartments by retractable doors. Prior to formal testing, rats underwent a 10-day habituation protocol. From days 1 to 7, animals were acclimated daily for 2 h to testing conditions (23 °C, 50 lux illumination). During days 8–10, graduated exposure was implemented, beginning with 3 min of central compartment exploration followed by 3 min of full arena access each day.

Experimental Sequence: All behavioral assessments were conducted between 09:00 and 12:00 during the animals' active phase. The test comprised two sequential phases:

Sociability Test (0–10 min): Following a 5 min baseline period with empty wire cages in both lateral chambers, an age-/sex-matched Sprague–Dawley "Stranger 1" rat was introduced into one lateral cage while the opposite chamber retained an empty control cage. Social exploration behavior (sniffing, following, physical contact) directed toward Stranger 1 vs. the empty cage was video-recorded.

Social Novelty Preference (10–20 min): A novel "Stranger 2" rat replaced the empty cage, while Stranger 1 remained in its original location. The time spent interacting with the familiar Stranger 1 versus the novel Stranger 2 was quantified to assess social novelty discrimination.

Rats were individually placed in clean polycarbonate cages under 50 lux illumination. After 10 min habituation, spontaneous grooming behaviors (facial cleaning, body licking, genital/tail grooming) were video-recorded for 10 min using a Logitech C920 HD Pro webcam (manufactured by Logitech International S.A., headquartered in Lausanne, Switzerland).

SD rats were individually placed in polycarbonate test chambers containing 5 cm of fresh corncob bedding. After a 3 min acclimatization period, animals were temporarily transferred to holding cages. Sixteen black glass marbles (1.6 mm diameter) were arranged in a standardized 4 × 4 grid pattern spaced 3 cm apart. Subjects were returned to the chamber for a 10 min testing session under 50 lux illumination. Sessions were video-recorded from orthogonal angles.

Behavioral assessments were conducted in a square open-field arena measuring 100 × 100 × 40 cm (L × W × H), which featured a clearly demarcated central zone (50 × 50 cm). Animal movement throughout the arena was continuously recorded using an overhead camera system for subsequent analysis. Following a 10 min habituation period during which animals were allowed to freely explore the novel environment, formal behavioral testing commenced for a duration of 10 min. Throughout this testing phase, we quantified specific exploratory and anxiety-related behaviors using the Noldus EthoVision XT software.

Spatial learning and memory were assessed in a Morris water maze consisting of a black circular pool (150 cm diameter) filled with temperature-controlled water (23.0 ± 0.5 °C), opacified using nontoxic white tempera paint. A hidden escape platform (15 cm diameter) was submerged 1.5 cm below the surface in the target quadrant. Four high-contrast geometric patterns provided consistent distal spatial cues under uniform illumination (50 ± 5 lux). Animal movements were tracked using an overhead camera system (EthoVision XT v15.0, Noldus Information Technology, Wageningen, The Netherlands) with positional accuracy validated at <2% error via chessboard calibration prior to testing.

During the 5-day acquisition phase, animals completed four daily trials (90 s maximum/trial) with 25 min inter-trial intervals. Starting points (north, south, east, west) were randomized across trials. Animals failing to locate the platform within 90 s were manually guided to it and remained there for 10 s for spatial orientation. Twenty-four hours after the final acquisition session, spatial memory was evaluated in a 60 s probe trial where the platform was removed. All behavioral parameters were quantified from video recordings using EthoVision XT software.

Animals were sacrificed after completion of behavioral tests. Anesthesia was induced by intraperitoneal administration of ketamine (100 mg/kg) and xylazine (10 mg/kg). For biochemical analyses, rats (n = 12 per group) were decapitated without perfusion. The brains were rapidly removed, and the prefrontal cortex and hippocampus were dissected on ice, snap-frozen in liquid nitrogen, and stored at −80 °C until assayed. For histological evaluations, rats (n = 6 per group) were transcardially perfused with 0.9% NaCl followed by fixation with 4% paraformaldehyde (PFA, Merck, Germany). After careful removal of the brain, it was post-fixed in 4% PFA for 24 h. The hippocampus and prefrontal cortex were subsequently isolated, processed, and paraffin-embedded.

The levels of oxidative stress markers (GSH, T-SOD, GSH-PX, MDA, CAT, T-NOS, NO) and pro-/anti-inflammatory cytokines (IL-6, IL-10, IL-1β, TNF-α) were quantified using commercial kits (Nanjing Jiancheng Institute of Biotechnology, Nanjing, China) following the manufacturers' guidelines.

Fresh fecal samples were collected from experimental rats using aseptic procedures after immobilization:

Tissue homogenization was performed with 0.1 mm zirconium beads (BioSpec Products, 11079101z, Bartlesville, OK, USA) under liquid nitrogen pre-cooling using a FastPrep-24 homogenizer (MP Biomedicals, Solon, OH, USA).

DNA purity criteria: NanoDrop 2000 (Thermo Fisher Scientific, Waltham, MA, USA) absorbance ratios A260/A280 = 1.8–1.9, A260/A230 > 2.0. Qualified DNA was diluted to 1 ng/μL using sterile ultrapure water (Millipore Milli-Q, Merck KGaA, Darmstadt, Germany).

The V3-V4 hypervariable regions were amplified using primers 341F (5′-CCTAYGGGRBGCASCAG-3′) and 806R (5′-GGACTACNNGGGTATCTAAT-3′).

PCR conditions:

Template: Diluted genomic DNA

Primers: Barcode-labeled specific primers

Enzyme: Phusion^® High-Fidelity PCR Master Mix with GC Buffer (New England Biolabs, Ipswich, MA, USA)

Thermal cycling: 98 °C for 1 min; 30 cycles (98 °C for 10 s, 50 °C for 30 s, 72 °C for 30 s); 72 °C for 5 min.

Amplified products (~550 bp) were verified by 2% agarose gel electrophoresis (1× TAE buffer) and purified using the TIANgel Midi Purification Kit (Tiangen, Beijing, China, DP209).

Libraries were prepared with the NEBNext Ultra II DNA Library Prep Kit (Illumina, San Diego, CA, USA, E7645S). Quality control included fragment distribution analysis (Agilent 5400 Bioanalyzer, Agilent Technologies, Santa Clara, CA, USA; target peak: 450–650 bp) and quantification (Qubit 4.0, Thermo Fisher; concentration >20 nM).

Sequencing was performed on the NovaSeq 6000 platform (Illumina, San Diego, CA, USA) using PE250 mode, with a depth of 50,000 reads/sample.

QIIME2 (v2023.2) pipeline: DADA2 denoising, chimera removal, and ASV table generation. Taxonomic annotation utilized the SILVA 138.1 database (V3-V4 region extraction).

α-diversity: Chao1, Shannon, and Faith's PD indices (QIIME2 core-metrics).

β-diversity: Bray–Curtis and UniFrac distances visualized via PCoA/NMDS.

Multi-method validation: LEfSe (LDA score > 2.0), ANCOM (W-statistic > 0.7), DESeq2 (FDR < 0.05).

Co-occurrence networks: Constructed in Gephi (v0.10.1; Spearman |ρ| > 0.6, p < 0.01).

Multivariate analysis: RDA (vegan package) and PLS-DA (mixOmics package) in R (v4.3.1) using the "vegan" (v2.6-4) and "mixOmics" (v6.24.0) packages, respectively.

Brain sections were deparaffinized and subjected to antigen retrieval. After three 5 min washes with phosphate-buffered saline (PBS, pH 7.4), non-specific binding sites were blocked with 4% (w/v) bovine serum albumin (BSA) in PBS containing 0.3% Triton X-100 (PBST) for 1 h at room temperature (RT). Sections were incubated with rabbit primary antibodies (1:200 dilution in blocking solution) targeting specific antigens at 4 °C for 16 h in a humidified chamber. Following PBST washes (3 × 10 min), samples were treated with Alexa Fluor 488-conjugated goat anti-rabbit IgG secondary antibodies in the dark at RT for 1 h. Nuclei were counterstained with 4′, 6-diamidino-2-phenylindole (DAPI; 1 μg/mL) for 10 min in darkness.

Confocal imaging was performed using a laser scanning microscope (Leica TCS SP8 STED 3X, Leica Microsystems, Wetzlar, Germany) with the following standardized parameters: Objective: 63× oil immersion (HC PL APO CS2; NA = 1.40); laser lines: 488 nm (Alexa Fluor 488, Thermo Fisher Scientific, Waltham, MA, USA; 15% power), 405 nm (DAPI; 8% power); detection windows: 500–550 nm (green channel), 410–480 nm (blue channel); pinhole: 1.0 Airy unit (optical section thickness: 0.8 μm); scan mode: Sequential line scanning (frame averaging: 3×; scan speed: 400 Hz); Z-stack: Step size 0.5 μm (total depth: 8–10 μm per stack); and image resolution: 1024 × 1024 pixels (pixel size: 120 nm). Detector gains were calibrated daily using control sections to avoid saturation (max pixel intensity <4095 AU). Z-stack images were processed for maximum intensity projection and analyzed for integrated density (IntDen) using ImageJ (Fiji v2.3.0) with background subtraction (rolling ball radius: 50 pixels).

Statistical analyses were performed using GraphPad Prism software (version 9.0; GraphPad Software, San Diego, CA, USA). Continuous data are presented throughout the figures and text as mean ± standard deviation (SD). Normality of data distribution was assessed using the Shapiro–Wilk test. For comparisons between the two experimental groups, an unpaired two-tailed Student's t-test was employed when data met assumptions of normality and homogeneity of variance. Where data violated normality assumptions, the non-parametric Mann–Whitney U test was used instead. Statistical significance was defined as a p-value less than 0.05 for all analyses.