Changes in gut bacteria may worsen brain inflammation and thinking problems in offspring exposed to prenatal VPA

All experimental protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of the Guangzhou Institute of Biomedicine and Health, Chinese Academy of Sciences (Approval No. CAS [IACUC:2023081]) and conducted in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, 8th edition, 2011). Procedures adhered to the 2020 AVMA Guidelines for Euthanasia and the ARRIVE 2.0 reporting guidelines (Percie du Sert et al., 2020). All efforts were made to minimize animal suffering, including the use of humane endpoints and appropriate analgesic regimens.

All behavioral assessments were conducted in a dedicated testing room under controlled conditions (23 ± 1°C, 50 lux illumination). Mice were habituated to the testing environment for 1 hour prior to each procedure. Experimenters were blinded to group assignments throughout data collection and analysis.

Following behavioral assessments, mice were deeply anesthetized via intraperitoneal injection of ketamine/xylazine (100/10 mg/kg) and transcardially perfused with ice-cold 0.9% NaCl followed by 4% paraformaldehyde (PFA; Merck, Germany). Brains were post-fixed in 4% PFA for 24 hr at 4°C, after which the hippocampus and prefrontal cortex were dissected. Tissues were dehydrated through graded ethanol series, embedded in paraffin (Merck), and sectioned coronally at 5 μm thickness using a rotary microtome (Microm HM335E, Walldorf, Germany). Sections were mounted on Superfrost Plus slides (Thermo Fisher) for subsequent analyses.

Oxidative stress markers (GSH, T-SOD, GSH-PX, MDA, CAT) and nitric oxide metabolites (T-NOS, NO) were quantified using commercial kits (Nanjing Jiancheng Institute of Biotechnology) following manufacturer protocols. Pro- and anti-inflammatory cytokines (IL-6, IL-1β, TNF-α, IL-10) were measured in serum using mouse-specific ELISA kits (Fankel, Shanghai) with inter- and intra-assay coefficients of variation <10%. Absorbance was read on a SpectraMax M5 microplate reader (Molecular Devices).

Sample Collection: Fresh fecal pellets (3–5 pellets/mouse, ~100 mg) were collected aseptically between 09:00–11:00 to minimize circadian variability. Samples were snap-frozen in liquid nitrogen within 5 min and stored at −80°C until processing.

Brain sections underwent standard deparaffinization and antigen retrieval protocols. After three washes (5 min each) with phosphate-buffered saline (PBS, pH 7.4), nonspecific binding was blocked by incubating sections 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 then incubated with rabbit primary antibodies (1:200 dilution in blocking solution) targeting specific antigens at 4°C for 16 h under humidified conditions. Following PBST washes (3 × 10 min), samples were incubated with Alexa Fluor 488-conjugated goat anti-rabbit IgG secondary antibodies in the dark for 1 h at RT. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; 1 μg/mL) for 10 min in the dark.

High-resolution z-stack images were captured using a confocal laser scanning microscope (TCS SP8 STED 3X, Leica Microsystems) with standardized laser power and detector gain settings. Integrated density (IntDen) values were quantified using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Statistical analyses were performed using GraphPad Prism software. Continuous data are presented throughout the figures and text as mean ±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.

The open field test was conducted to evaluate autonomous exploratory behaviors and anxiety-like responses in C57BL/6 mice exposed to a novel environment. Valproic acid (VPA) exposure induced significant reductions in exploratory activity and exacerbated anxiety-like behaviors compared to control mice (Figure 1). Key findings are detailed below:

The three-chamber test evaluated social motivation (0–10 min) and social novelty preference (10–20 min) in C57BL/6 mice by quantifying exploration time toward unfamiliar conspecifics (Stranger 1/2) versus empty cages/objects.

I. Social Motivation Phase (Sociability, 0–10 min)

VPA-exposed mice exhibited markedly reduced social motivation compared to controls (Figure 2A):

(1) Spatial Exploration Preference

Stranger 1 cage duration: VPA-exposed mice spent significantly less time in the Stranger 1 cage than controls (P < 0.0001).

Empty cage duration: VPA-exposed mice showed a pronounced preference for the empty cage over the Stranger 1 cage (P < 0.0001).

(2) Social Interaction Behavior

Stranger 1 sniffing time: Interaction durations with Stranger 1 were significantly shorter in VPA-exposed mice than in controls (P < 0.0001).

Object sniffing time: VPA-exposed mice also displayed reduced interaction times with objects compared to controls (P < 0.05).

Representative locomotor trajectories during the sociability phase are shown for the ASD model group (Figure 2C) and control group (Figure 2D).

Conclusion: VPA-exposed mice avoided interactions with unfamiliar conspecifics (Stranger 1) and preferentially engaged with nonsocial contexts (empty cage), indicating impaired social competence (P < 0.0001).

II. Social Novelty Preference Phase (10–20 min;Figure 2B)

(1) Spatial Exploration Patterns

Stranger 1 cage duration: VPA-exposed mice spent significantly more time in the Stranger 1 cage than controls (P < 0.0001).

Stranger 2 cage duration: Interaction time with the novel Stranger 2 cage was significantly reduced in VPA-exposed mice compared to controls (P < 0.0001).

(2) Social Interaction Disparities

Stranger 1 sniffing time: VPA-exposed mice exhibited increased interaction with Stranger 1 compared to controls (P < 0.01).

Stranger 2 sniffing time: Exploration of the novel social stimulus (Stranger 2) was significantly diminished in VPA-exposed mice (P < 0.0001).

Representative trajectories for the novelty preference phase are shown for the ASD model group (Figure 2E) and control group (Figure 2F).

Conclusion: VPA-exposed mice displayed restricted interest in social novelty and failed to exhibit a clear preference for novel stimuli.

The Morris water maze (MWM) was employed to evaluate spatial learning and memory in C57BL/6 mice through two phases: spatial acquisition trials and spatial probe trials.

ELISAs demonstrated that valproic acid (VPA) exposure significantly altered the inflammatory cytokine profile in the prefrontal cortex of C57BL/6 mice (Figure 4). Key results are summarized below:

ELISAs revealed that valproic acid (VPA) exposure significantly disrupted oxidative stress homeostasis in the prefrontal cortex of C57BL/6 mice (Figure 4). Specific findings are detailed below:

To evaluate neuroinflammatory pathology in VPA-exposed C57BL/6 mice, immunofluorescence staining was performed to assess microglial activation via the marker Iba1.

In the hippocampal CA1 region, Iba1 fluorescence intensity was significantly higher in VPA-exposed mice than in controls (P < 0.0001; Figures 5A–G). Similarly, in the prefrontal cortex, Iba1 fluorescence intensity was markedly elevated in VPA-exposed mice compared to controls (P < 0.0001; Figures 5H–N). These findings demonstrate that VPA exposure significantly increased microglial Iba1 expression in both the hippocampal CA1 region and prefrontal cortex relative to normal controls.

Conclusion: VPA exposure triggered microglial activation, shifting microglia from a resting state to a proinflammatory phenotype. This suggests neuroinflammatory mechanisms may contribute to ASD-related behavioral abnormalities.

Comparative analysis of gut microbiota across taxonomic hierarchies revealed distinct compositional differences between VPA-exposed mice and controls (Figures 6A–F).

Class level (Figure 6A): Clostridiales was enriched in the VPA group, suggesting dysbiosis favoring taxa linked to gut-brain axis modulation.

Family level (Figure 6B): Muribaculaceae (associated with mucosal homeostasis) was reduced in the VPA group.

Genus level (Figure 6C): Oscillibacter (butyrate-producing genus) was diminished in the VPA group.

Order level (Figure 6D): Prevotellaceae_NK3B31_group (linked to carbohydrate metabolism and short-chain fatty acid production) was reduced in the VPA group.

Phylum level (Figure 6E): Bacteroidia (critical for polysaccharide degradation) decreased in the VPA group, while Clostridia (pro-inflammatory taxa) increased.

Venn diagram (Figure 6F): Substantial overlap in operational taxonomic units (OTUs) between groups was observed, alongside unique OTU profiles indicating conserved and divergent microbial niches.

These taxonomic shifts reflect gut microbiota dysbiosis in the VPA-induced ASD model. Reduced Bacteroidia suggest altered metabolic and inflammatory pathways, while diminished Oscillibacter and Muribaculaceae may indicate impaired butyrate synthesis and mucosal barrier function. Conversely, the control group increased Prevotellaceae_NK3B31_group suggest adaptive microbial shifts potentially influencing neuroactive metabolite production.

The depletion of Bacteroidia, essential for dietary fiber fermentation, may compromise gut barrier integrity and anti-inflammatory metabolite synthesis. Reduced Oscillibacter (a butyrate producer) might disrupt gut-brain axis energy homeostasis. These findings highlight the gut microbiota as a dynamic interface in ASD pathophysiology, warranting further investigation into its role in neurobehavioral outcomes.

Heatmap analysis revealed distinct taxonomic shifts in gut microbiota between VPA-exposed mice and controls across multiple taxonomic levels (Figures 7A–D).

Family level (Figure 7A): Lachnospiraceae (short-chain fatty acid [SCFA] producers) was elevated in the Control group.

Genus level (Figure 7B): Prevotellaceae UCG_001 (implicated in carbohydrate metabolism) was unenriched in the VPA group. Increased abundances of Ruminococcaceae (fiber-degrading taxa) and Lactobacillaceae (beneficial commensals) were also observed, suggesting compensatory mechanisms against dysbiosis.

Order level (Figure 7C): Enterobacterales, a taxon linked to pro-inflammatory responses and metabolic dysregulation, showed increased abundance in the VPA group.

Phylum level (Figure 7D): Pseudomonadota, a phylum associated with neurotoxin production, was enriched in the VPA group.

These shifts indicate a dysbiotic gut microbiota state in the VPA-induced ASD model, characterized by elevated taxa associated with neurotoxicity (e.g., Pseudomonadota), inflammation (e.g., Enterobacterales), and altered SCFA metabolism (e.g., Lachnospiraceae). The coexistence of harmful and compensatory taxa highlights the complexity of gut-brain interactions in ASD.

Enrichment of pro-inflammatory (Enterobacterales) and neurotoxin-producing (Pseudomonadota) taxa may synergistically disrupt gut barrier integrity, amplify systemic inflammation, and perturb neural signaling. These findings implicate gut microbiota as a potential modulator of ASD phenotypes, warranting further investigation into causal mechanisms.

LEfSe analysis revealed hierarchical dysbiosis in the VPA-exposed mice compared to controls across taxonomic levels (Figures 8A–F).

The gut microbiota composition of VPA-exposed mice exhibited significant taxonomic shifts compared to controls, with consistent trends in pro-inflammatory and beneficial bacterial taxa (Figures 9A–D).

Multiple alpha diversity metrics revealed reduced microbial richness and evenness in VPA-exposed mice compared to controls (Figures 10A–F).

Species richness: Controls showed significantly higher Chao1 (Figure 10A), Faith’s Phylogenetic Diversity (Faith PD; Figure 10B), and observed features (Figure 10C) indices than VPA-exposed mice, indicating diminished microbial richness in the ASD model.

Diversity indices: The Shannon index (species richness and evenness; Figure 10D) was markedly lower in VPA-exposed mice, while the Simpson index (taxonomic dominance; Figure 10E) was diminished in the ASD group.

Rarefaction analysis: The Shannon rarefaction curve (Figure 10F) plateaued at higher diversity levels in controls, reflecting a more stable and taxonomically rich microbiome compared to the ASD group.

VPA-exposed mice exhibited global gut microbiota dysbiosis, characterized by:

Reduced richness (fewer unique taxa);Uneven species distribution (dominance of specific taxa);Loss of phylogenetically distinct lineages (lower Faith PD), potentially impairing metabolic and immunomodulatory functions.

These patterns align with clinical ASD studies, where diminished alpha diversity correlates with impaired microbial functional redundancy and ecosystem resilience.

Beta diversity analyses across multiple distance metrics demonstrated pronounced structural divergence in gut microbiota between the VPA-induced ASD mouse model group and the normal control group, with distinct intra- and inter-group heterogeneity patterns (Figures 11A–F).

Using Bray-Curtis dissimilarity (Figures 11A, B), the ASD model group exhibited significantly greater intra-group distances compared to controls (Figure 11A), indicative of heightened compositional variability among ASD individuals. Conversely, inter-group comparisons (Figure 11B) revealed distinct clustering between ASD and control cohorts despite reduced absolute distances. Similar trends were observed with unweighted UniFrac analysis (Figures 11C, D), where ASD mice displayed expanded phylogenetic diversity within the group (Figure 11C) but maintained significant separation from controls (Figure 11D). Weighted UniFrac analysis (Figures 11E, F) further corroborated these findings, showing both increased intra-group dispersion in the ASD model (Figure 11E) and robust inter-group differentiation (Figure 11F), particularly in taxa abundance-weighted phylogenetic space.

Collectively, these results highlight two key features of gut microbiota in the ASD model: 1) elevated intra-group heterogeneity, suggesting microbial community instability or individualized dysbiosis patterns, and 2) persistent inter-group segregation, reflecting conserved structural shifts distinguishing ASD-associated microbiota from healthy configurations. The concordance across dissimilarity metrics (Bray-Curtis, unweighted/weighted UniFrac) underscores the multidimensional nature of these differences, spanning compositional, phylogenetic, and abundance-driven microbial characteristics.

Phylogenetic heatmap profiling revealed systematic shifts in gut microbiota composition between the VPA-induced ASD mouse model group and the normal control group, with distinct taxonomic signatures (Figures 12A, B).

Phylogenetic Tree Heatmap with Taxonomic IDs (Figure 12A):The ASD model group exhibited marked depletion of Turicibacter, Prevotellaceae_NK3B31_group, and Prevotellaceae_UCG_001, alongside Lactobacillus (Figure 12A).

Phylogenetic Tree Heatmap (Figure 12B):Phylum-level analysis demonstrated divergent trends between groups (Figure 12B). The VPA group showed elevated representation of Thermodesulfobacteriota and Pseudomonadota, phyla often linked to sulfate metabolism and oxidative stress regulation, respectively. In contrast, the ASD model group displayed reduced Verrucomicrobiota (primarily Akkermansia), a phylum critical for mucin layer homeostasis, alongside diminished Actinomycetota.

The network analysis(Figures 13A–D) revealed a complex interaction pattern among different bacterial families. The nodes representing bacterial families were color-coded based on their respective phyla, including Actinomycetota (red), Bacillota (yellow), Bacteroidota (green), Patescibacteria (blue), Pseudomonadota (orange), Thermodesulfobacteriota (purple), and Verrucomicrobiota (pink). The size of the nodes indicated the relative abundance of each bacterial family, while the thickness of the edges reflected the strength of the interactions between them.

Our results highlight the intricate relationships within the microbial community. Notably, certain phyla such as Actinomycetota and Bacteroidota exhibited frequent interactions, suggesting a potential co-dependency or synergistic relationship. In contrast, other phyla like Patescibacteria and Pseudomonadota showed fewer interactions, indicating possible niche specialization or competitive exclusion. These findings provide valuable insights into the dynamics of microbial communities and their ecological roles in various environments.

Functional annotation of microbial pathways across KEGG hierarchical levels (L1-L3) and MetaCyc databases demonstrated systematic metabolic shifts in the VPA-induced ASD mouse model group compared to the normal control group, characterized by coordinated dysregulation of energy and neuroactive metabolite-related pathways (Figures 14A–D).

The VPA-treated group showed a slightly higher proportion in the Metabolism category (Figure 14A), particularly in Energy metabolism and Amino acid metabolism (Figure 14B). Additionally, the Alanine, aspartate and glutamate metabolism pathway (Figure 14C) and the ANAGLYCOLYSIS-PWY pathway (Figure 14D) exhibited increased proportions in the VPA-treated group compared to the control group.

Collectively, these functional shifts delineate a microbial metabolic landscape in ASD models marked by: Energy metabolism imbalance: Reduced carbohydrate utilization, suggesting inefficient energy harvesting. Neurotransmitter precursor dysregulation: Divergent trends in glutamate- and glycine-related pathways align with ASD-associated excitatory/inhibitory signaling imbalances. The hierarchical consistency of these changes—from broad metabolic categories (L1) to substrate-specific pathways (L3)—strengthens the biological plausibility of gut microbiota metabolic dysfunction as a feature of the VPA-induced ASD model.