Gut Bacteria From Autistic Children Reshape a Key Brain Protein in Mice

If a child's autism traces back to the trillions of microbes living in their gut, scientists want to know exactly how. A new mouse study offers one of the more detailed looks yet: transplanting fecal bacteria from children with autism spectrum disorder (ASD) into lab mice didn't just change the animals' behavior, it altered the chemical switches on proteins deep in their brains. The researchers trace much of that change to a single protein, SHANK3, that appears to lose its grip on a synaptic partner once a gut-driven chemical switch gets flipped.

Key takeaways

• Mice given fecal bacteria from children with ASD developed autism-like behaviors (less interest in unfamiliar mice, more repetitive marble-burying, less exploring of open space) alongside changes to hundreds of brain protein "switches."
• A computer model of SHANK3, a synaptic scaffolding protein already linked to autism, predicted a much weaker grip on its usual partner protein once one specific switch was flipped on.
• FNDC3A, a related protein, was also elevated in the blood plasma of the children with ASD themselves, the only part of the mechanism confirmed outside mice.
• The core mechanism rests on a mouse transplant experiment plus computer structure predictions, not on direct evidence from human brain tissue.

Gut Bacteria From Children With Autism Altered Brain Protein Switches in Mice

Researchers took fecal samples from 20 children diagnosed with ASD and 20 typically developing children, then transplanted pooled samples into two groups of antibiotic-treated mice (15 animals each) over three weeks of daily dosing, a technique called fecal microbiota transplantation (FMT, moving gut bacteria from one body into another). The design builds on an influential 2019 mouse study that first showed ASD-associated gut bacteria alone could trigger autism-like behavior in mice, and this new work, published in the journal Microorganisms, pushes further into the underlying biochemistry.

Rather than looking only at which proteins the hippocampus (the brain region central to memory, and to social behavior in rodent autism models) made more or less of, the team mapped phosphorylation, the addition of a small phosphate tag that works like a chemical on/off switch for a protein. The results were lopsided: only 29 proteins differed meaningfully in raw abundance between the two groups of mice, but 479 phosphopeptides (out of nearly 20,000 measured) changed significantly. Gut microbiota from children with ASD appeared to rewire which switches were flipped on brain proteins far more than it changed how much of those proteins existed in the first place.

Why This Finding Matters for Families Navigating an Autism Diagnosis

Autism now affects roughly 1 in 31 8-year-olds in the United States, according to CDC surveillance data, and there is still no blood test or brain scan that can diagnose it; clinicians rely on behavioral criteria alone. That gap is part of why any molecular lead draws attention from families already managing ASD, IBS, IBD, or other conditions they suspect trace back to gut health.

Mice that received ASD-associated bacteria showed changes that echo features of autism in children: a clear preference deficit toward unfamiliar mice in social-novelty testing (a standard rodent measure of social interest), less movement and less exploring of open space, and more marbles buried, a proxy for repetitive, compulsive-like behavior. The mice also showed more active immune cells (microglia and astrocytes) in the hippocampus, fewer newly forming neurons, and physically thinner dendritic spines, the tiny protrusions where synapses form.

Inside the Study: Transplants, Phosphoproteomics, and AI Structure Modeling

Behavioral testing was only the starting point. The researchers ran mass-spectrometry-based proteomics and phosphoproteomics on hippocampal tissue from five mice per group, then used pathway analysis showing the affected phosphoproteins clustered heavily around synapse-related pathways. Two proteins stood out: SHANK3, a well-established autism risk gene that scaffolds the postsynaptic density (the dense cluster of proteins just inside a receiving synapse), and SRRM2, a splicing-related protein separately tied to developmental delay and autism features in a prior report of 22 patients.

To test whether the phosphorylation changes mattered structurally, the team turned to AlphaFold3 (AF3), the AI system described in Nature for predicting protein shapes, plus a docking program called HADDOCK that estimates how tightly two proteins bind. The modeling predicted that adding a phosphate tag to SHANK3 at one specific site drops its predicted binding strength to a key partner protein, Homer1, from -5.4 to -1.8 kcal/mol, a large computed loss in binding affinity. Separately, cross-referencing the proteomic and phosphoproteomic hits turned up FNDC3A, a protein previously linked to ASD through gene duplications, as elevated in both the mouse hippocampus and, notably, in blood plasma from the same children with ASD.

What This Could Mean for Someone Managing Autism Today

For a parent or adult reading this while researching a diagnosis, the honest translation is: not much yet, in practical terms. Nothing here validates a probiotic, diet change, or supplement as a treatment, and the authors don't test one. The elevated FNDC3A levels in blood are a research signal, not an approved biomarker; no clinical test is built on this finding. What the study does offer is a sharper question for future research: whether interventions targeting abnormal SHANK3 phosphorylation specifically, rather than broad microbiome changes, might eventually be worth exploring. That's a hypothesis to watch, not something to raise with a doctor yet.

The Catch: Mouse Models and Computer Predictions Are Not Proof

The study's own authors flag the central limitation plainly, calling the AlphaFold3 and HADDOCK results "predictive working hypotheses," not measured structures. Nothing here confirms, with lab techniques like NMR or cryo-electron microscopy, that SHANK3 actually folds differently inside a living neuron, in a mouse or a person. FNDC3A is the one part of the chain checked against real human tissue (blood plasma from 20 children with ASD versus 20 controls); the SHANK3/SRRM2 phosphorylation and structural story was built entirely in mice and in silico. The authors also note their donor pool came from just five fecal samples per group, risking one atypical donor skewing the whole transplant, and that findings from this all-male cohort may not generalize to the broader, highly heterogeneous autism population.

What Comes Next for This Research

The researchers call for direct experimental confirmation: structural techniques such as cryo-electron microscopy to check whether SHANK3's predicted shape change is real, and mutagenesis experiments that swap out the specific phosphorylation site to see if it actually weakens the protein's binding in a dish. For FNDC3A, they say knockdown and overexpression studies are needed to show whether raising or lowering the protein actually changes behavior, rather than just tracking alongside it. Until those experiments land, this remains a mechanistic hypothesis from one mouse-transplant study, not a confirmed pathway.

Common Questions

Does this study prove gut bacteria cause autism? No. It shows that transplanting gut bacteria from children with ASD into mice produced autism-like behaviors and brain changes in those mice. It does not establish the same causal chain in humans, and the one piece confirmed in people, elevated FNDC3A in blood, stops well short of proving causation.

What is SHANK3? SHANK3 is a scaffolding protein that helps hold together the molecular machinery at synapses, the junctions where neurons communicate. Mutations in the SHANK3 gene are already a well-established cause of some autism cases, including Phelan-McDermid syndrome.

Is there a blood test for autism based on FNDC3A now? No. FNDC3A was elevated in blood plasma from children with ASD in this one study, but it has not been validated as a diagnostic test and is not used clinically.

Why did phosphorylation change so much more than overall protein levels? The researchers found only 29 proteins whose total abundance meaningfully differed between groups, but 479 phosphopeptides changed significantly. That suggests gut bacteria from children with ASD may act mainly by flipping molecular switches on existing proteins rather than by changing how much of each protein the brain makes.

References

1. Yongsheng Ge, Zhi Li, Caiyun Yu, et al. (2026). Gut Microbiota-Elicited Aberrant Phosphorylation Induces Protein Structural Anomalies: A Non-Negligible Pathogenic Driver of Autism Spectrum Disorder. Microorganisms. https://doi.org/10.3390/microorganisms14091925
2. Autism Data Visualization Tool. cdc.gov. https://www.cdc.gov/autism/data-research/autism-data-visualization-tool.html
3. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature. https://www.nature.com/articles/s41586-024-07487-w