AI Clocks Point to an Oral Microbe That Slows Aging in Humans, Worms and Mice

A common bacterium that lives quietly on the human tongue may hold one of the most surprising clues yet in the search for interventions that slow aging. In a study published in Nature Aging, a team led by Jing-Dong J. Han at Peking University reports that Neisseria flavescens, an oral commensal long regarded as an unremarkable resident of the mouth, is consistently associated with decelerated biological aging in people, and that supplementing animals with this microbe extends lifespan in worms and partially reverses molecular signs of aging in mice. The finding elevates the oral microbiome, long overshadowed by its gut counterpart, into a serious candidate axis for geroprotective therapy.

The discovery did not begin with a hypothesis about any particular bacterium. Instead, the researchers built AURORA, a generative artificial intelligence framework that integrates multiple data modalities, including metagenomic sequences, plasma metabolomics, gene expression and physiological measurements, into a unified model of human aging. Within this framework, the team constructed multi-modality aging clocks, statistical models that estimate a person’s biological age from molecular and phenotypic features, and then used the model to perform in silico screening: systematically simulating thousands of potential interventions to identify which changes would most effectively reduce the gap between predicted biological age and chronological age.

That computational screen produced an unexpected front-runner. When the abundance of Neisseria flavescens was virtually increased in the model, the predicted age gap shifted in a favorable direction, and the simulated perturbation also produced beneficial changes in physiological signatures, promoted the abundance of known health-associated gut taxa and enhanced the biosynthesis of metabolites with documented benefits. In a natural human aging cohort, individuals classified as slow agers, whose biological clocks ran younger than their chronological years, carried significantly more of this oral species than fast agers. The association was replicated in longitudinal data tracking one-year changes in bacterial abundance against changes in the aging measure.

To move beyond correlation, the team isolated two live strains of N. flavescens, designated a11 and e5, from human donors and sequenced their genomes, which have been deposited under the accessions GWHHOEZ01000000 and GWHHOEA01000000. Laboratory analysis confirmed that both strains actively produce vitamins and other beneficial metabolites predicted by the computational framework. Targeted mass spectrometry of bacterial culture supernatants verified the metabolic output, providing a mechanistic bridge between the human association data and the biological activity of the organism itself.

The first functional test took place in Caenorhabditis elegans, the transparent roundworm that has long served as a workhorse of aging research. When worms were fed live N. flavescens instead of their standard laboratory diet of Escherichia coli OP50, their lifespans lengthened and their healthspans, the portion of life spent in good condition, improved. Supplementation also attenuated the age-related decline in mobility during early adulthood and altered body size, with independent replication experiments confirming the lifespan extension for both strains. Heat-killed bacteria likewise preserved worm mobility, hinting that structural components of the microbe, rather than only its living metabolic activity, may contribute to the effect.

In aged mice, the researchers turned to heat-killed N. flavescens, a form of supplementation that avoids the safety considerations of administering a live organism. After ten weeks of treatment, the animals’ serum metabolome, liver transcriptome and gut microbiome all shifted measurably toward younger profiles. Principal-component analyses showed that the metabolic and transcriptional states of treated twenty-month-old mice moved closer to those of young two-month-old controls, and regression analysis demonstrated that treatment-associated fold changes across metabolites, microbial species, microbial pathways and liver transcripts tended to oppose the changes normally driven by aging itself. The effect was partial rather than a wholesale reversal, but the consistent directional rescue across three distinct molecular layers is striking.

The study also addressed how an oral bacterium could plausibly influence systemic aging. Paired analyses of oral and gut microbiome data, including paired samples from the Human Microbiome Project, revealed links between N. flavescens abundance and the composition of the gut ecosystem, including associations with Akkermansia muciniphila, a gut bacterium widely studied for its beneficial effects on metabolic health and barrier function. Predicted downstream effects included changes in immune signaling pathways in peripheral blood mononuclear cells, suggesting a route by which oral microbial signals propagate through host metabolism and immunity rather than acting locally in the mouth alone.

Technically, the work represents a notable advance in how candidate interventions are identified. Traditional microbiome-aging studies typically search for correlations between individual taxa and chronological age or survival, an approach that the authors note would not have singled out N. flavescens as a top hit in their own data. By instead simulating perturbations within a generative model trained on multi-omics data, AURORA can prioritize organisms whose expansion is predicted to shift the entire aging state favorably, a logic closer to a causal intervention than a population association. The framework and all analysis code are freely available, and the underlying sequencing datasets have been deposited in public repositories under BioProject PRJCA057750 and associated accessions.

Important caveats remain. The human evidence is associative, the mouse experiments used heat-killed preparations whose active components have not yet been identified, and no human intervention trial has been conducted. A patent covering N. flavescens and its applications is pending, filed by two of the authors, which signals commercial interest but also that translational development is still at an early stage. Even so, the convergence of computational prediction, human cohort data, worm lifespan experiments and mouse molecular rescue makes a compelling case that the mouth, a microbiome site routinely dismissed as merely a gateway to dental disease, deserves far more attention in geroscience. If the findings hold up in further studies, the next geroprotective intervention might not come from a laboratory-synthesized molecule but from a common commensal already living between our teeth.

Subject of Research: Identification of the oral commensal Neisseria flavescens as a geroprotective microbe linked to decelerated human aging

Article Title: Multi-modality profiling identifies Neisseria flavescens as a central geroprotective oral commensal in humans

Article References: Chen, J., Ren, Y., Zhou, Y., Wang, Z., Li, J., Guo, X., Xu, H., Wang, Y., Tang, H., & Han, J.-D. J. (2026). Multi-modality profiling identifies Neisseria flavescens as a central geroprotective oral commensal in humans. Nature Aging. https://doi.org/10.1038/s43587-026-01220-0

Image Credits: AI Generated

DOI: 10.1038/s43587-026-01220-0

Keywords: aging, oral microbiome, Neisseria flavescens, AURORA, geroprotection, aging clocks, multi-omics, Caenorhabditis elegans, gut microbiome, probiotics, longevity, metabolomics