Sirtuins, and SIRT6 in particular, have long been considered candidates for longevity regulators, but their role as biomarkers of aging remains disputed. Earlier work relied mostly on individual model organisms and did not always account for confounding factors such as secondary mutations or the FUDR effect. A large body of data has already accumulated around this topic, from studies of epigenetic clocks to analyses of methylation and cellular landscapes of aging. This work sits at the intersection of those directions, but stands out by its scale: it is the first to systematically test SIRT6 against a large corpus of literature and transcriptomic data from different species.
The authors collected 1648 article abstracts on SIRT6 from OpenAlex and Europe PMC and applied BERTopic topic modeling, including a zero-shot mode with predefined topics based on hallmarks of aging. They then built the SIRT6.db database from 54 transcriptomic datasets across six organisms (human, mouse, rat, macaque, pig, Drosophila) with SIRT6 knockout or overexpression. To search for conserved signatures, they used meta-regression and a rank-based approach, along with analysis of transcription factor activity. In addition, they studied SIRT6 expression in human tissues from ARCHS4 and in single-nucleus data on Alzheimer's disease.
Topic modeling showed that SIRT6 is most often linked to cellular senescence (199 articles), mitochondrial dysfunction (171), and telomere shortening (136). The meta-analysis revealed that SIRT6 knockout causes changes that go in different directions depending on species and tissue, but mitochondrial dysfunction turned out to be the most conserved response. Among genes, only ZNF518A (up) and BZW1 (down) changed consistently. The activity of the transcription factors NF-κB, AP-1, and STAT differed between mouse and human. In human tissues, SIRT6 expression mostly did not change with age, but in skin fibroblasts it fell after age 80, accompanied by rising variability. In Alzheimer's disease, SIRT6 decreased in neurons already at early stages, but increased in microglia, while in astrocytes it fell only at late stages.
The work shows that SIRT6 is not a universal biomarker of chronological aging, and that its effects depend on context: species, tissue, sex, and cell type. This casts doubt on simple schemes in which SIRT6 directly controls aging. Unlike neighboring studies that focus on individual organisms or data types, this one is the first to combine a literature corpus with cross-species transcriptomic data, which makes it possible to see the limitations of existing models. The results underscore the need to account for cellular composition and disease stage when interpreting SIRT6 changes.
Further studies could test how post-translational modifications and the localization of SIRT6 affect its functions in different tissues. Studying the role of SIRT6 in proteostasis, which so far has been little explored in this context, looks promising. It is also important to conduct longitudinal studies in humans with repeated measurements, in order to assess the dynamics of SIRT6 over time and its link to functional decline.
The results will influence the development of biomarkers of aging and the interpretation of sirtuin data in geroprotective research. They also matter for neuroscience, since they reveal cell-specific changes in SIRT6 in Alzheimer's disease.
The next step could be validating the signatures found (ZNF518A, BZW1) in independent cohorts and at the protein level. It would also be worth running experiments with partial SIRT6 knockout, to separate the effects of complete loss from dose-dependent ones.
The work is directly tied to the problem of reproducibility in aging biomarkers: most candidates fail to hold up on independent data. It also touches on a fundamental question, namely which mechanisms of aging are conserved between species and which are specific to particular tissues or diseases.
In skin fibroblasts from people over 80, SIRT6 expression lands in the top 2.4% of genes with the greatest increase in variability, meaning that in some long-lived individuals it is high and in others low, which feels a lot like a lottery.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.