Researchers at MIT, Massachusetts General Hospital and partner institutions have linked optical signs of cell chemistry to gene activity in aging mouse tissue. Their method, RamanOmics, helps identify senescent cells, which have stopped dividing but remain active. Measuring the chemistry with light leaves the sample available for further analysis, allowing researchers to compare different kinds of evidence from the same cells.
Finding these cells is difficult because senescence can look different in different tissues. A molecular marker that works in one setting may miss cells in another. Raman microscopy adds a view of their chemistry: it shines light on a sample and measures the small changes in the light's energy caused by vibrating chemical bonds. The resulting pattern, called a spectrum, carries information about molecules such as lipids and proteins without requiring a chemical label.
To connect those patterns to cell activity, the team first scanned mouse lung and skin tissue with Raman microscopy. It then used STARmap, a technique that maps which genes are active and where, on the same tissue sections. Separate measurements of RNA from cell nuclei supplied a broader gene-activity reference. The researchers used this reference to estimate additional gene activity in the mapped cells, linking chemical measurements to a fuller picture of what those cells were doing.
The combined information helped a machine-learning model distinguish cells showing p21, a marker associated with senescence, from cells without that signal. In the mouse-lung test, it correctly classified about 78% of cells when given both Raman and gene-activity features, compared with 74% using gene activity alone. Performance also improved in skin. The team trained on 70% of the selected cells and tested on the remainder, after balancing the two groups. These were comparisons within the study, rather than tests on independent patients.
That result shows the value of combining the measurements, but does not establish that a light scan alone can reliably identify every senescent cell. The study used p21 to define the group of interest, which covers only part of the varied senescent population. Raman signals also overlap: a change associated with lipids does not necessarily identify one exact compound or show that it caused the cell to become senescent.
For now, RamanOmics gives researchers a way to examine how a cell's chemistry relates to its gene activity while retaining its location in tissue. Turning that approach into a practical diagnostic remains a substantial step. MIT reports that the current system takes about 30 hours to scan one square millimetre of tissue, and the reported findings come from mice.
Sources
- Nature Aging: RamanOmics decodes the spatial vibrational–molecular architecture of senescence in aging and repair
- MIT News: Unmasking zombie cells in aging tissue with an AI-powered barcode
- Raman spectrometer photograph by IOM Leipzig, CC BY 4.0. Representative laboratory instrument; it is not the apparatus used in the study.