Prebiotic diet normalizes aberrant immune and behavioral phenotypes in a mouse model of autism spectrum disorder

ASD is characterized by social communication deficits, cognitive impairments, and repetitive behaviors, with rising global prevalence and higher incidence in males. Emerging evidence implicates gut-brain axis dysfunction, including altered gut microbiota and increased gastrointestinal (GI) problems, in ASD pathogenesis. Microbiota from individuals with ASD can transfer behavioral phenotypes to rodents, underscoring causality. Prebiotics, non-digestible oligosaccharides that modulate beneficial gut bacteria and produce SCFAs, may improve epithelial barrier integrity, mucosal immune homeostasis, serotonergic signaling, neuroinflammation, and behavior. This study asked whether early-life dietary supplementation with specific prebiotics (3% GOS/FOS 9:1) can mitigate behavioral, immune, neuroinflammatory, and microbiome alterations induced by prenatal VPA exposure in mice, with a focus on immune mediation of gut-brain effects.

The introduction summarizes evidence that individuals with ASD have increased GI complaints and altered gut microbiota composition, with severity of GI issues correlating with behavioral severity. Fecal microbiota transplantation from ASD donors induces ASD-like behaviors in rodents. Prebiotics can increase beneficial bacteria, yield SCFAs that strengthen epithelial tight junctions and immune balance, and modulate serotonergic systems, neuroinflammation, cognition, and sociability. However, clinical and preclinical evidence is still emerging, necessitating further mechanistic and interventional studies focusing on microbiome-based therapies in ASD.

Design: Prenatal exposure VPA mouse model with early-life prebiotic dietary intervention. BALB/cByJ mice (8-week-old males n=8, females n=16) were bred; gestational day (G) 0 at vaginal plug. On G11, dams received VPA 600 mg/kg s.c. or PBS. Diets (AIN-93G control or AIN-93G + 3.0% GOS:FOS 9:1) were provided to lactating dams from postnatal day (P)0 through weaning (P21) and continued in offspring thereafter. Groups: Male offspring: (1) PBS control diet, (2) PBS GOS/FOS, (3) VPA control diet, (4) VPA GOS/FOS. Behavioral testing: Social interaction (P31, P49) using a two-chamber open field with perforated cages; sociability index SI = (time with target − time with empty)/(time with target + time with empty). Spatial memory (P30, P48) in an object location paradigm across 5 sessions; discrimination index DI = (Δtime with displaced objects)/(Δtime with non-displaced). Two breeding cohorts (n=25 and n=13). Exact sample sizes per test reported in figures (e.g., PBS control n=7–9; VPA control n=9–10; PBS GOS/FOS n=10; VPA GOS/FOS n=9 for behavior). Tissue collection: At P50, mice were anesthetized and decapitated; cecum, ileum, spleen, blood, and brain collected. Neuroinflammation assays: Microglia morphology in mPFC via iDISCO clearing and light-sheet imaging (Iba1+ cell number, volume, occupancy). Cerebellar (CRB) microglial activation markers by Western blot: TMEM119, CD68 (normalized to GAPDH). Serotonergic system: Left hemisphere iDISCO for serotonergic neurons (Tph2+) in raphe nuclei (RN): positions along rostro-caudal axis and counts. Intestinal assays: Ileal enterochromaffin (5-HT+) cell counts by IHC; tight junction protein ZO-1 by immunofluorescence with corrected total fluorescence quantification. Peripheral immunity: Splenic T helper subsets by flow cytometry: Th1 (CD69+Tbet+), Th17 (CCR6+RorγT+), Treg (CD25+FoxP3+) within CD4+; ratios (Th1+Th17)/Treg. Serum haptoglobin by ELISA. Microbiome: Cecal 16S rRNA gene V4 sequencing (Illumina MiSeq, 2×250); OTU clustering at 97% against Greengenes in QIIME 1.8; rarefaction to 1000 reads/sample. Diversity: Shannon and reciprocal Simpson (Phyloseq). Beta diversity: Bray–Curtis PCoA; ADONIS (vegan). Differential taxa: LEfSe (alpha 0.05, LDA > 2.0). SCFAs (acetic, propionic, butyric, isobutyric, isovaleric, valeric) quantified by GC with 2-ethylbutyric internal standard. Statistics: Two-way ANOVA for effects of exposure (VPA vs PBS), diet (GOS/FOS vs control), and interaction, with Tukey’s post hoc; data transformations for non-normal variables; outliers removed by robust regression and outlier removal; ADONIS on distance matrices (999 permutations); Spearman correlations between SCFAs and taxa. Significance P<0.05.

- Behavior: VPA controls lacked social preference at P31 and P49. GOS/FOS completely restored sociability in VPA mice at both P31 (P<0.05) and P49 (P<0.0001). Sociability index (SI) was reduced by VPA at both time points (P<0.0001) and increased by GOS/FOS (P<0.0001 at P31; P<0.05 at P49). VPA impaired spatial discrimination at P30 and P48 (P<0.0001 vs PBS controls). GOS/FOS improved spatial recognition in VPA mice at P48 (P<0.01), not at P30. - Neuroinflammation: In cerebellum, VPA increased TMEM119 expression overall [F(1,20)=5.917], indicating microglial activation; diet had no main effect on TMEM119. GOS/FOS reduced CD68 expression overall [F(1,20)=5.646], with a significant decrease in VPA GOS/FOS vs VPA control (P<0.05). No significant mPFC microglial differences by exposure or diet (Iba1+ number, volume, occupancy). - Serotonergic neurons: VPA reduced Tph2+ RN neuron number [F(1,20)=24.31, P<0.0001]; GOS/FOS significantly mitigated this loss in VPA mice (P<0.01). No effect on RN positional distribution. - Peripheral immunity: Th1 proportion unchanged by VPA (P=0.35) but decreased by GOS/FOS in VPA-exposed mice (P<0.05). Th17 increased with VPA [F(1,20)=7.79, P<0.05]; GOS/FOS reduced Th17 in VPA mice (P<0.05). Treg unchanged by main effects but exposure×diet interaction significant [F(1,20)=13.46, P<0.01], with a trend to higher Treg in VPA GOS/FOS vs PBS GOS/FOS (P=0.0504). The (Th1+Th17)/Treg ratio was higher in VPA control vs PBS control (P<0.01) and reduced by GOS/FOS in VPA mice (P<0.01). - Serum haptoglobin: VPA decreased HP; significant exposure×diet interaction [F(1,21)=12.84, P<0.01]; GOS/FOS restored HP in VPA mice (P<0.05). - Intestinal measures: EC 5-HT+ cells in ileum showed exposure×diet interaction [F(1,18)=8.90, P<0.01]; GOS/FOS increased 5-HT+ cells in VPA mice vs VPA controls (P<0.05). ZO-1 expression decreased with VPA [F(1,19)=19.96, P<0.001] and increased with diet [F(1,19)=4.97, P<0.05]; GOS/FOS restored ZO-1 in VPA mice (P<0.01), indicating improved barrier integrity. - Microbiome diversity and composition: Diet reduced alpha diversity (Shannon: [F(1,20)=16.72, P<0.001]; Simpson: [F(1,20)=26.72, P<0.0001]); VPA had an overall effect on Simpson [F(1,20)=6.52, P<0.05]. Bacteroidetes/Firmicutes ratio showed exposure and exposure×diet effects; GOS/FOS increased B/F ratio in VPA mice (P<0.0001) but not PBS. Beta diversity showed significant group clustering (ADONIS R²=0.5354, P<0.001); diet explained substantial variance (R²=0.2421, P<0.001). - Differential taxa and SCFAs: LEfSe identified taxa distinguishing PBS vs VPA; GOS/FOS restored 8/9 characteristic control taxa in VPA mice (all but Ruminococcaceae). SCFAs: VPA affected acetic, propionic, isobutyric, valeric; diet affected propionic, isobutyric, isovaleric, valeric. In VPA GOS/FOS vs VPA control: propionic increased (P<0.0001), acetic decreased (P<0.01), isovaleric and valeric increased (P<0.001, P<0.01). Correlations: propionic positively correlated with Prevotella (r=0.51), but acetic negatively with Prevotella (r=-0.56); butyric positively with Bacteroidales (r=0.46), negatively with Ruminococcus (r=-0.49); acetic positively with Clostridiales (r=0.67); Peptococcaceae positively with branched SCFAs; Ruminococcaceae negatively with propionic and positively with branched SCFAs.

The study demonstrates that early-life supplementation with GOS/FOS prebiotics normalizes key ASD-like phenotypes in male mice prenatally exposed to VPA, including sociability and spatial memory deficits. These behavioral improvements coincide with restoration of gut barrier integrity (ZO-1), increased intestinal 5-HT-producing enterochromaffin cells, normalization of peripheral immune balance (reduced Th17, lowered (Th1+Th17)/Treg ratio), partial attenuation of cerebellar microglial phagocytic activation (reduced CD68), recovery of serotonergic neuron numbers in the raphe nuclei, and rebalancing of gut microbiota composition, including restoration of key taxa and SCFA profiles. Together, findings support a mechanistic link along the microbiota–gut–immune–brain axis, where prebiotics modulate microbial ecology and metabolites, improve mucosal and systemic immune homeostasis, and reduce neuroinflammation to ameliorate ASD-like behaviors. The region-specific brain effects (cerebellum vs mPFC) suggest targeted neuroimmune modulation. The data highlight the potential of nutritional interventions to influence neurodevelopmental outcomes through immune and microbial pathways.

Prenatal VPA exposure disrupts gut microbiota composition and metabolism, increases gut permeability, skews peripheral T cell balance toward pro-inflammatory states, induces cerebellar microglial activation, reduces serotonergic neurons, and impairs social behavior and spatial memory. Early-life dietary supplementation with 3% GOS/FOS (9:1) reverses many of these alterations: it restores key bacterial taxa and SCFA patterns, improves intestinal barrier and EC 5-HT+ cell numbers, normalizes systemic immune balance and serum haptoglobin, reduces cerebellar phagocytic microglial activation, rescues serotonergic neuron counts, and corrects behavioral deficits. These results underscore the therapeutic potential of prebiotics to modulate the gut–immune–brain axis in ASD. Future research should include assessment of repetitive behaviors, direct measurements of brain 5-HT neurotransmission, functional assays of gut permeability, evaluation of sex differences by including females, and broader translational studies to determine efficacy and mechanisms in clinical populations.

- Only male mice were studied; females were excluded because they do not display the ASD-like phenotype in this VPA model, limiting generalizability across sexes. - The VPA-induced ASD model does not capture the full heterogeneity of human ASD and has inherent translational limitations. - Repetitive behaviors (a core ASD symptom) were not assessed to reduce testing burden; this limits behavioral scope. - No direct measurements of 5-HT concentrations in brain regions were performed to confirm neurotransmission changes. - Intestinal barrier function was inferred from ZO-1 expression without functional permeability assays. - Brain neuroinflammation assessments showed region-specific effects; broader CNS mapping was limited by sample sizes per region.