Moscow State University preprint: some mammalian DNA methylation sites follow the same U-shaped age trajectory as mortality, and aging clocks built on this signal outperform existing ones at detecting blood rejuvenation and Parkinson's disease
Moscow State University preprint: some mammalian DNA methylation sites follow the same U-shaped age trajectory as mortality, and aging clocks built on this signal outperform existing ones at detecting blood rejuvenation and Parkinson's disease
On September 14, Stanislav Tikhonov and Sergei Dmitriev of Lomonosov Moscow State University published a preprint comparing blood DNA methylation across 16 human chronic diseases with aging and mortality data from 42 mammalian species. In the four species for which infant methylation data were available (human, cat, zebra, and bottlenose dolphin), between a quarter and a half of significant sites changed with age in the same shape as that species' mortality curve.
All-cause mortality does not increase linearly with age: it is high immediately after birth, falls to a minimum around ages 8 to 10, and rises again through the rest of life. This pattern is called the U-shaped curve. The authors asked whether this curve leaves a trace in DNA methylation, the set of chemical marks on the genome that keep some genes switched on and others switched off.
Aging is conventionally described as a monotonic accumulation of cellular damage, but all-cause mortality follows a U-shaped trajectory with age, which means that some molecular changes should be nonmonotonic as well.
The idea has a precedent: the activity of the IGF1 gene, which drives body growth during childhood, first rises and then falls. To test whether many epigenomic sites repeat this pattern and to separate residual developmental programming from aging, Tikhonov and Dmitriev compared methylation in adult chronic diseases with infant methylation data and species mortality curves.
Across species, disorder in DNA methylation on average increased without reversal, consistent with the damage-accumulation model. But in the same four species, a fraction of sites behaved differently, tracking the shape of the mortality curve: 24.8% in humans, 43.2% in zebras, 49.8% in bottlenose dolphins, and 50.2% in cats. One such site lies near the ASGR1 gene, a target of cholesterol-lowering drugs; another lies near NFIX, which is required for muscle and blood cell formation. Both genes are active in blood throughout life, yet the methylation marks near them trace the bend of the mortality curve.
In the chronic disease dataset, Tikhonov and Dmitriev found two opposing groups: some diseases increased methylation disorder in blood, while others decreased it. In Huntington's disease, blood methylation disorder fell even though the disease destroys the brain; in mice and sheep, blood disorder declined while in mouse brain it rose. The authors interpret this as a compensatory response: blood maintains order while another organ is under stress.
Using this signal, they built a clock that predicts mortality risk across multiple species and tissues. In mice and sheep, the clock outperformed existing cross-species age clocks, and in human blood it matched the accuracy of specialized clocks such as Horvath, Hannum, PhenoAge, and DunedinPACE, even though eight such clocks applied to the same blood sample diverged by an average of 17 years. The new clock also responded more sensitively to heterochronic parabiosis (surgical joining of the circulatory systems of an old and a young mouse), partial reprogramming, and exercise in mice. In human blood, it was the only clock to detect Parkinson's disease; it also detected osteoporosis, but only after childhood mortality was included in the training data, confirming that the advantage came specifically from the U-shaped signal.
Tikhonov and Dmitriev connect this finding to Mikhail Blagosklonny's hyperfunction theory of aging: part of age-related pathology is produced by growth and developmental programs that the organism fails to shut down after the body has finished forming.