MIT identifies a shared chemical fingerprint of cellular aging: a lipid signal identical in mouse lung and skin yields a precise barcode for detecting senescent cells without destroying tissue
MIT identifies a shared chemical fingerprint of cellular aging: a lipid signal identical in mouse lung and skin yields a precise barcode for detecting senescent cells without destroying tissue
On September 21, 2026, a team from MIT and Massachusetts General Hospital published a method called RamanOmics in Nature Aging. It combines label-free Raman microscopy (in which light scatters off chemical bonds within a cell without damaging it) with RNA sequencing performed on the same mouse tissue sections. In lung and skin, the method found different sets of active genes but the same chemical signal of aging.
With age, some cells enter senescence: they stop dividing but do not die, and they continue to influence their neighbors through inflammation and tissue remodeling. These cells are typically sought using the protein markers p21 and p16, but neither one identifies a senescent cell unambiguously, because each marker also appears in cells of different tissues and at different stages. p21 and p16 mostly label distinct populations. Removing p21-positive cells prevents radiation-induced bone loss in mice, so the authors chose p21 as the primary marker. This summer, the U.S. National Institutes of Health proposed describing a senescent cell by a set of coordinates that includes chemical composition. That is precisely the measurement RamanOmics delivers for the first time.
MIT showed in 2024 that a Raman spectrum of a living cell can predict its transcriptional profile. Now Ke Zhang, Salvatore Sorrentino, and colleagues have linked that relationship to RNA sequencing and the STARmap spatial gene-activity map, working on sections of lung and skin from mice aged two and 26 months (barrier organs in contact with the environment). In old mice, lung tissue upregulated inflammation and tissue-repair genes, while skin upregulated keratinization genes: the gene programs diverged. Yet in both tissues the same Raman peak at 1131–1135 cm⁻¹ intensified, a signature of a specific class of lipids. This is the first shared aging marker found across organs whose genetic signatures differ.
From the most informative Raman peaks and genes, the authors assembled a barcode and trained a classifier to distinguish senescent cells from normal ones more accurately than any single signal alone: accuracy rose from 73.7% to 77.6% for lung and from 61.8% to 65.5% for skin. Validation on a mouse skin wound ruled out coincidence: on the third day after injury, p21 activity increased, and with it the same keratinization genes and the same lipid peak observed in aging.
"By combining the most important Raman features with the most important genetic signatures, we were able to create a barcode that helps find senescent cells more objectively," says MIT postdoc Salvatore Sorrentino in an MIT News article.
Jeon Woong Kang, also from MIT, describes the ultimate goal:
"You can imagine that one day we could develop an endoscope capable of looking inside the body and recognizing cellular senescence."
For now, the method works only in mice, and analyzing a tissue area of about 1 mm² takes roughly 30 hours. Senescent cells are rare even in aged tissue, accounting for fractions of a percent of all cells. The finding at this stage is correlational: the causal role of lipids will be tested by knocking out the enzymes responsible for elongating them.
Because the aging signal resides in scattered light rather than in RNA (whose extraction requires destroying the cell), the method can in principle be accelerated and adapted to living tissue. This opens a path toward nondestructive monitoring of senescent cell burden and toward verifying whether senolytics, drugs designed to selectively clear such cells, actually reduce it.