Salk Institute researchers have created a molecular map of how healthy mice respond to exercise, offering clues that could help improve performance and reduce injury risk.
Exercise can help the body resist disease and dysfunction and recover from illness. Studies have also linked it to more than 30 percent lower mortality from all causes, cardiovascular disease, and cancer. Explaining those benefits is harder, partly because studies in people generally rely on blood and muscle samples, leaving researchers with an incomplete picture of how exercise changes other organs.
Researchers at the Salk Institute turned to mice to follow those changes through more of the body. They compared young male and female mice that ran voluntarily on a wheel for eight weeks with mice that remained sedentary, creating a first-of-its-kind molecular atlas spanning 22 tissues. The study was published in Cell on October 15, 2026.
No single gene responded everywhere
Exercise altered the activity of nearly half the genes examined in at least one tissue, yet no single gene’s activity changed in all 22. Those comparisons came from bulk RNA sequencing, which measures gene activity across whole tissue samples. Rather than revealing one response shared everywhere, the analysis showed that tissues adapted in different ways.
The researchers also cataloged the small molecules that cells use and produce during metabolism in eight tissues and plasma, the liquid part of blood. Alongside this metabolomics analysis, they used proteomics to measure protein levels in four specific leg muscles, examining a hamstring muscle, two calf muscles, and the quadriceps.
“Exercise doesn’t just change our muscles, it changes many of our organs and cells at different levels—some short-term, some long-term,” says co-first and co-corresponding author Geraldine Maier, PhD, a staff scientist at Salk. “Our experimental setup allows us to capture that detail across tissues, and sexes, and in muscle, down to the individual cell types.”
Training changes DNA marks in muscle
Skeletal muscle accounted for most of the changes in gene activity, and the strong responses in leg muscles prompted the researchers to examine three of them more closely. Using single-nucleus RNA sequencing, they could distinguish neighboring cell types and compare their gene activity instead of averaging their responses together. They also conducted the first single-cell analysis of DNA methylation in muscle after exercise.
In some muscle cell types, training changed chemical marks on DNA that help regulate gene activity, while other cell types showed no such changes. These marks are added through a process called DNA methylation. Changes in their pattern can be more persistent than shifts in RNA or protein levels, allowing this form of epigenetic regulation to have lasting effects on how particular cell types and whole tissues behave.
“Exercise impacts the whole body, not just your muscles, so an atlas like ours that looks more holistically at many tissues is key to understanding the true impact of exercise,” says co-first author Manoj Hariharan, PhD, a former senior staff scientist at Salk. “That’s also why single-cell studies are so important—there are so many cell type-specific changes that get lost in bulk sequencing. I hope our work highlights that and encourages more single-cell atlases, and especially multi-omic single-cell atlases, moving forward.”
Exercise responses vary by sex and time
Female mice showed broader changes in gene activity and distinct expression patterns compared with males. The epigenetic response to training also differed strongly by sex.
To capture variation across daily biological rhythms, the researchers collected tissues during both the active and resting phases of the mice’s circadian cycle. Most tissues showed greater exercise-related changes when sampled during the active phase, with the hypothalamus, kidney, and lung as exceptions.
“Our atlas provides a comprehensive resource for understanding the systemic molecular adaptations to endurance exercise,” says co-first author Ramesh Ramasamy, PhD, a former computational biologist at Salk. “Now that we have all this data, researchers can begin developing strategies that harness exercise—either alone or in combination with other interventions—to ensure we all live long, healthy lives.”
51 genes distinguish trained muscle
The researchers have already used the gene activity dataset to develop a 51-gene exercise signature, a pattern that distinguishes exercise-trained muscle from sedentary muscle in both mice and humans. It could help identify biomarkers, measurable indicators of a biological state, and support research into muscle disuse and dysfunction.
By establishing a healthy baseline, the atlas gives researchers a reference for investigating how tissues adapt to exercise. The differences by sex and circadian phase could eventually inform more personalized exercise routines, while closer investigation of particular tissues and cell types may point to ways of improving performance, recovery, and muscle health.
“Now that we have the data,” adds Maier, “labs can identify one gene or metabolic pathway they find interesting, then dissect that mechanism. Our atlas gives the field a shared foundation to build on.”
How long those molecular changes last remains unresolved. Researchers still need to determine how long training must continue before lasting changes develop and whether cells retain a memory of exercise for months or years.
Reference: “Multi-omics atlas of chronic exercise” by Geraldine Maier, Manoj Hariharan, Ramesh K. Ramasamy, Aidan Glina, Laura van Rosmalen, Suraj J. Pathak, Shaunak Deota, Sahil Patel, Terry Lin, Michael Tun Yin Lam, Hiep D. Le, Krushali Powale, Takashi Shimomura, Yuta Otobe, Vince Rothenberg, Ming Sun, Takeshi Sakurai, Tiffany Chin, Yue Wu, Rosa Castanon, Anna Bartlett, Huaming Chen, Joseph R. Nery, Cesar Barragan, Swetha Devulapalli, Hemal H. Patel, Alan Saghatelian, Hikari Yoshitane, Debashis Sahoo, Susan M. Keach, Joseph R. Ecker and Satchidananda Panda, 15 October 2026, Cell.
DOI: 10.1016/j.cell.2026.09.044
The work was supported by the Wu Tsai Human Performance Alliance; Joe and Clara Tsai Foundation; National Institutes of Health (S10 OD026929, P30 014195, HG010634); Chapman Foundation; Helmsley Charitable Trust; Ministry of Education, Culture, Sports, Science and Technology in Japan; Veterans Administration; National Science Foundation (2138259, 2138286, 2138307, 2137603, 2138296); and Howard Hughes Medical Institute.
Never miss a breakthrough: Join the SciTechDaily newsletter.
Follow us on Google and Google News.