Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Oct 28;13(11):922.
doi: 10.3390/toxics13110922.

Investigating the Effects of Long-Term Fine Particulate Matter Exposure on Autism Spectrum Disorder Severity: Evidence from Multiple Analytical Approaches

Jianrui Dou  1 , Kaiyue Zhang  1 , Ruijin Xie  2 , Hua Xu  2 , Qiyang Pan  3 , Xue Xiao  2 , Yufan Luo  2 , Shengjie Xu  2 , Wei Xiao  2 , Dongqin Wu  2 , Bing Wang  1 , Linpei Zhang  4 , Chenyu Sun  5   6   7 , Yueying Liu  2
Affiliations

Investigating the Effects of Long-Term Fine Particulate Matter Exposure on Autism Spectrum Disorder Severity: Evidence from Multiple Analytical Approaches

Jianrui Dou et al. Toxics. .

Abstract

With rapid industrial expansion, air pollution's adverse neurological effects have gained increasing attention. Children face a greater risk of neurological damage because of their higher breathing rates, developing brains, and limited ability to detoxify harmful substances. Fine particulate matter has been identified as a primary neurotoxic contributor affecting developing brains. Strong evidence connects environmental pollutant exposure to the prevalence of Autism Spectrum Disorder (ASD), a neurodevelopmental condition marked by lasting difficulties with social communication and interaction. This study explores the association between long-term PM2.5 exposure and ASD symptom exacerbation, investigating underlying mechanisms. We hypothesize that long-term PM2.5 exposure exacerbates ASD symptoms through neuroinflammatory activation, leading to neuronal damage and impaired synaptic plasticity. Our investigation employs three complementary approaches: First, integrated analysis combining Global Burden of Disease data with Mendelian randomization demonstrates a significant association between PM2.5 exposure and increased ASD severity risk. Second, utilizing the China High-Resolution Air Pollution Database in conjunction with cohort studies, we provide evidence that ambient air pollution substantially influences autism severity, with PM2.5 identified as the predominant environmental determinant. Third, through network toxicology, single-cell transcriptomics, and animal experimentation, we demonstrate that chronic PM2.5 exposure exacerbates valproic acid-induced autism-like behaviors in murine models, identifying CTNNB1, PTEN, CCR2, AKT1, and mTOR as potential core mediating genes. Importantly, these findings represent preliminary results. Several potential confounding factors such as co-exposure to other pollutants and socioeconomic variables have not been fully addressed. While our multi-modal approach provides converging lines of evidence, further validation in larger, more diverse populations with refined control of confounders will be essential to establish causality and elucidate mechanisms. Nonetheless, these early insights advance our understanding of PM2.5-induced neurotoxicity in the context of ASD and offer timely, albeit preliminary, evidence to inform public health policy.

Keywords: PM2.5; autism; co-exposure; network toxicology; neurotoxicity.

PubMed Disclaimer

Conflict of interest statement

All authors confirm that there are no conflicts of interest. Illustrations were created with BioRender.com.

Figures

Figure 1
Figure 1
Association Between Global Autism Spectrum Disorder (ASD) Burden and Air Pollution. (A) World map showing the age-standardized Disability-Adjusted Life Year (DALY) rate of Autism Spectrum Disorder per 100,000 population (ages 0–14), based on Global Burden of Disease 2021 data. (B) Spearman correlation analysis between country/regional PM2.5 concentrations and age-standardized DALY rates for ASD. (C) Mendelian Randomization of the forest plot evaluating the potential relationship between air pollution exposures (PM2.5, PM2.5–PM10, PM10, and NOx) and ASD risk. (D) Scatter plot of Mendelian Randomization instrumental variables, illustrating the potential positive effect of PM2.5 exposure on ASD symptoms. (E) Gene Ontology (GO) molecular function enrichment analysis based on single nucleotide polymorphisms (SNPs) used as instrumental variables in the Mendelian Randomization analysis using SNPnexus database. (F) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis based on SNPs used as instrumental variables in the Mendelian Randomization analysis using SNPnexus database(https://www.snpnexus.org/v4/, accessed on 25 September 2025).
Figure 2
Figure 2
Association Between Exposure to Air Pollution Mixtures and Autism Severity (CARS Scores) Using Multiple Co-Exposure Statistical Models. (A) Univariate exposure-response curves generated by Bayesian Kernel Machine Regression (BKMR), illustrating the independent and potentially non-linear associations of individual air pollutants with Childhood Autism Rating Scale (CARS) scores. The solid blue line represents the estimated effect, with the shaded gray area denoting the 95% credible interval. (B) Overall effect of the combined air pollution mixture on CARS scores, as estimated by the BKMR model. (C) Weighted Quantile Sum (WQS) regression results, showing the estimated weights of individual pollutants within the mixture index. These weights reflect the relative contribution of each pollutant to the overall positive association with CARS scores, withPM2.5 contributing the most. (D) Component weights from the Quantile g-Computation model (qgcomp), indicating the proportional contribution of each air pollutant to the total mixture effect. All pollutants exhibit positive weights, suggesting a consistent association with higher CARS scores.
Figure 3
Figure 3
Human Network Toxicology Analysis of Potential Gene Targets Linking PM2.5 Exposure and Autism. (A) Flowchart outlining the methodological approach for the human network toxicology analysis. (B) Venn diagram showing the overlap between PM2.5-related genes and autism-related genes. (C) Bar plot presenting GO molecular function enrichment results for the intersecting genes. (D) Bar plot illustrating KEGG pathway enrichment results for the intersecting genes. (E) Visualization of the protein–protein interaction (PPI) network constructed from the intersecting gene set. (F) Hub gene analysis identifying top-ranked hub genes using two algorithms: Maximal Clique Centrality (MCC) and Degree. (G) Venn diagram displaying the 14 overlapping hub genes identified by both MCC and Degree algorithms: AKT3, AKT1, BDNF, CCR2, CTNNB1, EGF, FOS, IGF1, KRAS, MAPK3, mTOR, PTEN, PXDN, and SOX2.
Figure 4
Figure 4
Bioinformatic and Machine Learning Analysis of Gene Expression in Autism Spectrum Disorder (ASD) Based on the GEO database. (A) Principal Component Analysis (PCA) of gene expression profiles from ASD patients (n = 170, red) and control subjects (n = 115, blue), demonstrating clear separation between the two groups. (B) Volcano plot depicting differentially expressed genes (DEGs) between ASD and control groups. Genes significantly upregulated in ASD are highlighted in red, while significantly downregulated genes are shown in blue. (C) GO enrichment analysis of DEGs, highlighting biological processes of the ATP hydrolysis in ASD. (D) KEGG pathway enrichment analysis of DEGs, indicating involvement of key pathways such as cAMP signaling and cell cycle regulation in ASD. (E) Architecture of the multi-layer perceptron neural network model for classifying ASD and control samples. The input layer includes selected gene features, with the top 10 most influential genes (including CTNNB1, PTEN, CCR2, AKT1, and mTOR) highlighted. (F) Boxplots comparing the relative expression levels of five key genes (CTNNB1, PTEN, CCR2, AKT1, and mTOR) between ASD and control groups. Each dot represents an individual sample, confirming significant differential expression of these genes in ASD. * p < 0.05, ** p < 0.01.
Figure 5
Figure 5
Single-cell RNA Sequencing Analysis of the Prefrontal Cortex from Individuals with Autism Spectrum Disorder (ASD) Based on the DISCO Database. (A) Uniform Manifold Approximation and Projection (UMAP) plots visualizing cell clusters in the prefrontal cortex from control (left) and ASD (right) samples. Each point represents a single cell, colored by cell type. Red arrows highlight significant differences in excitatory and inhibitory neuron clusters between the two groups. (B) Stacked bar charts showing the relative percentage of each cell type across individual control (Con1–Con3) and ASD (ASD1–ASD3) samples. (C) Stacked bar chart comparing the overall cell type distribution as a percentage between the aggregated control and ASD groups. (D) Stacked bar charts showing the absolute number of cells for each cell type recovered from individual control and ASD samples. (E) Stacked bar chart comparing the total absolute counts of each cell type between the aggregated control and ASD groups. ** p < 0.01.
Figure 6
Figure 6
Single-Cell Expression of Key Genes and Functional Enrichment Analysis in the Prefrontal Cortex of ASD. (A–E) Visualization of the expression of five key genes (CCR2, CTNNB1, PTEN, AKT1, and mTOR) at the single-cell level. For each gene, the UMAP plots (left) display the spatial distribution of gene expression across all cell types in control and ASD samples (higher expression shown with more intense red). The violin plots (right) compare the overall expression levels of each gene between control (blue) and ASD (red) groups. (F) GO enrichment analysis of differentially expressed genes between ASD and control samples (Top 30 terms). (G) KEGG pathway enrichment analysis of differentially expressed genes between ASD and control samples (Top 30 pathways). * p < 0.05, ** p < 0.01.
Figure 7
Figure 7
Effects of sustained environmental Ambient PM2.5 Exposure on Valproic Acid (VPA)-Induced Autism-Like Behaviors in Mice. (A) Flowchart illustrating the main steps for ambient PM2.5 sampling and collection. (B) Schematic of the animal experimental design, including prenatal VPA exposure and subsequent PM2.5 treatments. (C) Body weight curves of mice from four experimental groups over the 90-day exposure period. * indicate days on which body weight significantly differed from the control group (p < 0.05); & denote days with a significant difference between the ASD + Low PM2.5 and ASD + High PM2.5 groups (p < 0.05). (D) Diagram of the three-chamber social interaction test, showing the stages of habituation, sociability, and social novelty. (E) Results of chamber transitions during Stage 0. (F) Results from Stage 1 (Sociability Test), including relative sniffing duration (left) and sociability preference index (right) for each group. (G) Results from Stage 2 (Social Novelty Test), including relative sniffing duration (left) and social novelty preference index (right) for each group. (H) Novel Object Recognition (NOR) test results, showing exploration time for familiar (I) and novel (N) objects in each group. (I) Boxplot of the discrimination rate from the NOR test, reflecting cognitive and memory performance across the four groups. * p < 0.05, ** p < 0.01; N.S., not significant.
Figure 8
Figure 8
Effects of sustained environmentalAmbient PM2.5 Exposure on Valproic Acid (VPA)-Induced Autism-Like Behaviors in Mice. (A) Representative heatmaps from the open-field test, illustrating locomotor activity and exploration patterns across all experimental groups. (B) Quantification of the number of entries into the central zone of the open-field arena. (C) Quantification of the total time spent in the central zone during the open-field test. (D) Representative images from the marble-burying test across all groups. (E) Quantification of the percentage of marbles buried by mice in each group during the test. (F) Relative mRNA expression levels of five selected genes (CTNNB1, PTEN, CCR2, AKT1, and mTOR) in the prefrontal cortex of mouse brain tissue across all groups, as measured by RT-qPCR. * p < 0.05, ** p < 0.01.
See this image and copyright information in PMC

References

    1. Han Y., Yu Z., Chen Y., Guo X., Liu Y., Zhang H., Li Z., Chen L. PM2.5 induces developmental neurotoxicity in cortical organoids. Environ. Pollut. 2024;361:124913. doi: 10.1016/j.envpol.2024.124913. - DOI - PubMed
    1. Liu F., Liu C., Liu Y., Wang J., Wang Y., Yan B. Neurotoxicity of the air-borne particles: From molecular events to human diseases. J. Hazard. Mater. 2023;457:131827. doi: 10.1016/j.jhazmat.2023.131827. - DOI - PubMed
    1. Liu Q., Zhang D., Liang F., Liu F., Xiao L., An X., Chen X., Liang X. Air pollution and hypertension in rural versus urban children: Lipidomic insights into PM2.5 impacts. Environ. Res. 2025;278:121715. doi: 10.1016/j.envres.2025.121715. - DOI - PubMed
    1. Marsal A., Slama R., Lyon-Caen S., Borlaza L.J.S., Jaffrezo J.L., Boudier A., Darfeuil S., Elazzouzi R., Gioria Y., Lepeule J., et al. Prenatal Exposure to PM2.5 Oxidative Potential and Lung Function in Infants and Preschool- Age Children: A Prospective Study. Environ. Health Perspect. 2023;131:17004. doi: 10.1289/EHP11155. - DOI - PMC - PubMed
    1. Thiankhaw K., Chattipakorn N., Chattipakorn S.C. PM2.5 exposure in association with AD-related neuropathology and cognitive outcomes. Environ. Pollut. 2022;292 Pt A:118320. doi: 10.1016/j.envpol.2021.118320. - DOI - PubMed

LinkOut - more resources