Roundup’s Active Ingredient Reshapes Gut Bacteria and Disrupts Social Behavior, Even at ‘Safe’ Doses

Glyphosate, the active ingredient in Roundup, has long been framed as safe for mammals because it targets the shikimate pathway, a metabolic route mammals lack. That argument, however, says nothing about the trillions of gut microbes that do use this pathway. A new study published in Molecular Psychiatry by Matsuzaki R and colleagues finds that even at doses within the range regulators consider acceptable, glyphosate measurably remodels gut microbial communities and produces behavioral changes in mice, with a microbiota transplantation experiment pointing to a causal, not merely correlational, relationship.

Glyphosate at Acceptable Daily Intake Levels Disrupts Mouse Social Behavior

Male mice exposed to glyphosate at 50 mg/kg/day, the high-dose condition, lost their preference for a novel social partner in the Three-Chamber Test, a standard assay for social recognition. At both the lowest dose (0.5 mg/kg/day, which corresponds to the acceptable daily intake, or ADI) and the highest dose, male mice also spent significantly less time in the center of the Open Field arena, a reliable signal of heightened anxiety. Neither body weight nor plasma corticosterone levels, a hormonal marker of physiological stress, differed meaningfully between exposed and control animals, suggesting the effects were behavioral rather than a reflection of systemic toxicity. Female mice showed a different profile: locomotion dropped at the high dose, but the social and anxiety-like changes seen in males did not emerge clearly, pointing to sex-specific vulnerability. Prior work on glyphosate and gut-brain signaling had flagged neurobehavioral alterations as a concern, but most earlier studies used doses far above regulatory thresholds, making this study's focus on ADI-level exposure particularly relevant.

Why These Findings Matter for Human Exposure Assessment

Glyphosate is not a trace contaminant encountered by the occasional farmworker. Roughly 60 to 80 percent of the general population, including children, carry detectable glyphosate in bodily fluids, making it one of the most ubiquitous synthetic chemicals in the human body. Regulatory agencies set acceptable daily intakes on the assumption that the shikimate pathway, which glyphosate blocks, is irrelevant to mammalian biology. The new findings from Matsuzaki et al. complicate that logic: the gut microbiome is a mammalian organ system in every functional sense, and it is heavily populated by bacteria that depend on the shikimate pathway. Separately, childhood glyphosate exposure has been linked to liver inflammation and markers of metabolic disorder in young adulthood, raising the possibility that early gut-microbiome disruption could carry consequences that only become visible years later. The current study does not establish that same link in humans, but it strengthens the biological plausibility that current regulatory models may be missing a meaningful pathway of harm.

How a Transplantation Experiment Established a Causal Gut-Brain Link

The most methodologically significant piece of the study was not the behavioral testing itself but what came after. The team collected cecal microbiota from glyphosate-exposed male mice and transplanted it into antibiotic-treated, glyphosate-naïve male recipients. Those recipient mice, which had never encountered glyphosate directly, subsequently showed the same social behavioral impairments seen in directly exposed animals. This transplantation step is the closest a preclinical study can come to isolating the microbiome as a causal variable rather than a passive bystander. The anxiety-like phenotype did not transfer, suggesting that at least two independent mechanisms are operating: one gut-mediated, one presumably involving direct neurological or immunological effects of glyphosate itself.

Microbiota profiling revealed that glyphosate shifted the community composition (beta diversity) in both sexes without changing overall diversity (alpha diversity). In males, Lactobacillus strains increased while Bifidobacterium and Shigella dysenteriae strains declined. Predicted functional analysis flagged increases in butyrate synthesis II and acetate synthesis III alongside a decrease in inositol synthesis. Transcriptomic profiling of the amygdala in directly exposed males found altered expression in gene clusters tied to social behavior and learning or memory, including upregulation of B2m and downregulation of Arc. Crucially, these amygdalar changes were not replicated in microbiota-transplant recipients, which reinforces the interpretation that the brain transcriptome effects require direct glyphosate exposure, not just the altered microbiome.

Key Caveats Before Extrapolating to Human Health

This is investigational-stage research, and several limitations deserve explicit attention. The study used C57BL/6 mice, and converting mouse dose-per-kilogram to a human equivalent is not straightforward given differences in metabolism, exposure routes, and gut physiology. Real-world human exposure is chronic and cumulative across dietary sources, not a controlled seven-week drinking-water protocol. The team also did not synchronize estrous cycles in female mice, which could have introduced hormonal variability that muddied the social behavior data for that group. Antibiotic pretreatment used to clear microbiota before transplantation may itself alter host physiology in ways that confound the transplant results. And the microbiome analysis captured bacteria, leaving archaea, viruses, and eukaryotic gut residents uncharacterized. Researchers in the broader pesticide-microbiota field have called for more rigorous causal studies and better standardization of microbiome sequencing methods before firm conclusions can be drawn, a point that applies here as well.

What This Study Argues for in Pesticide Risk Assessment

Matsuzaki and colleagues frame their work explicitly within what they call ecological psychiatry: the idea that environmental exposures shape mental health through biological intermediaries like the gut microbiome. The practical implication they argue for is that pesticide risk assessments should incorporate functional endpoints from the gut-brain axis, not just classical toxicological markers like body weight and organ morphology. Under current frameworks, glyphosate passes safety thresholds partly because no gross physiological harm is visible at regulatory doses. The data here suggest that a finer-grained lens, one that tracks microbial community function and behavioral readouts, might tell a different story. That is not the same as concluding glyphosate is unsafe for humans at current dietary exposure levels: the evidence does not yet support that claim. What it does support is that the question deserves a more sophisticated answer than existing risk frameworks are designed to provide.

Evidence rating

INVESTIGATIONAL — 78/100

Caveats:

• This is an animal study conducted in mice; findings may not directly translate to human health outcomes.
• Funding source was not disclosed by the authors.
• Behavioral effects were sex-specific and dose-dependent, and the mechanisms driving anxiety-like behavior appear to be independent of microbiota changes.
• Real-world human glyphosate exposure patterns, routes, and durations differ substantially from the controlled conditions used here.
• Results have not yet been independently replicated in a separate laboratory or population.

Controlled animal study exposing male and female C57BL/6 mice to regulatory-range glyphosate doses for 7 weeks, with cecal microbiota transplantation used to probe causality. Demonstrates association and partial causal evidence (via transplant) between glyphosate-remodeled microbiota and social behavioral deficits in male mice, but findings are limited to an animal model and have not been replicated in an independent cohort. Funding was not disclosed by the authors.

References

1. Rie Matsuzaki, Benjamin Valderrama, Gabriel S. S. Tofani, et al. (2026). Towards ecological psychiatry: The herbicide glyphosate disrupts behavior through microbiota-gut-brain axis. Molecular Psychiatry. https://doi.org/10.1038/s41380-026-03693-2
2. Jarosław Mazuryk, Katarzyna Klepacka, Włodzimierz Kutner, Piyush Sindhu Sharma (2024). Glyphosate: Impact on the microbiota-gut-brain axis and the immune-nervous system, and clinical cases of multiorgan toxicity. Ecotoxicology and Environmental Safety. https://doi.org/10.1016/j.ecoenv.2024.115965
3. Flavia Silvia Galli, Marta Mollari, Valentina Tassinari, et al. (2024). Overview of human health effects related to glyphosate exposure. Frontiers in Toxicology. https://doi.org/10.3389/ftox.2024.1474792
4. Childhood exposure to common herbicide may increase the risk of disease in young adulthood – UC Berkeley School of Public Health. publichealth.berkeley.edu. https://publichealth.berkeley.edu/news/new-research-shows-adverse-impacts-of-childhood-glyphosate-exposure/
5. Rie Matsuzaki, Eoin Gunnigle, Violette Geissen, Gerard Clarke, Jatin Nagpal, John F Cryan (2023). Pesticide exposure and the microbiota-gut-brain axis. The ISME Journal. https://doi.org/10.1038/s41396-023-01450-9