Bacterial genomic structural variations in children with autism serve as diagnostic biomarkers

Gut microbiota dysbiosis is linked to autism spectrum disorder (ASD) in children. However, the role of bacterial genomic structural variations (SVs) in ASD remains largely unexplored.

We aimed to identify bacterial SVs associated with ASD and explore their mechanistic role and clinical application.

We collected faecal metagenomes from 452 children (261 ASD, 191 neurotypical) across an in-house and seven public datasets. Using linear mixed-effects modelling, we identified ASD-associated SVs and compositional shifts and validated candidate SVs in humanised gut microbiome mice.

We identified 100 bacterial SVs significantly associated with ASD (p<0.05). These SVs were enriched in genes involved in critical biological processes, including ion and amino acid metabolism and bacterial growth regulation in ASD. In particular, we found important SVs in Bacteroides uniformis related to thiamine and iron metabolism. Moreover, SVs in Ruminococcus torques were associated with the MazF (endoribonuclease toxin) and MazE (antitoxin) system, a key regulator of pathobiont proliferation. Validation in humanised mouse models confirmed significant correlations between these SV signatures and ASD-like behaviours, such as reduced social interaction and increased repetitive behaviours. Both phylogeographically conserved and regionally restricted SVs showed strong associations with ASD. A diagnostic model combining nine SVs and three bacterial species achieved an area under the receiver operating characteristic curve of 81.1%, outperforming models based solely on variable SVs (79.1%), deletion SVs (75.2%) or bacterial species abundance alone (72.3%).

Our findings suggest the significant role of bacterial genomic SVs in ASD and highlight their potential as diagnostic biomarkers.