One 45 minute bike ride changed 3,007 proteins across blood, saliva, and urine. In blood alone, 40% of the measured proteome shifted. A new Nature Metabolism atlas maps where those changes come from and which ones may matter for disease. Researchers at the University of Copenhagen, led by Nigel Kurgan, Bente Klarlund Pedersen, and Atul Deshmukh, tackled a basic question. We know exercise protects against diabetes, dementia, and heart disease. Meeting the 150 minutes per week guideline is linked to roughly 33% lower mortality. But how do organs coordinate that response? The answer matters for a practical reason. It may help settle whether the best workout is a specific format or simply consistent movement. The leading idea is exerkines (molecules released during exercise that carry signals between organs). Most past studies only looked at blood, with limited coverage and few timepoints. This one sampled plasma, saliva, and urine. The discovery cohort was 25 healthy men, average age 29, who cycled for 45 minutes at about 75% of heart rate reserve. The team collected blood, saliva, and urine before exercise, then again at 0, 0.5, 1, 3, and 24 hours afterward (urine was only collected before, right after, and at 24 hours). Across the three fluids, they measured 10,128 proteins. To see whether the results held up, they ran a second study in 24 inactive adults, average age 55, 15 of them women. Each person completed steady aerobic exercise, high intensity interval training (HIIT), and resistance training in random order, with blood drawn before and after each session. Right after exercise, 3,007 of the 10,128 proteins (30%) had changed. Blood showed the biggest shift: 2,156 proteins, or 40% of the plasma proteome (the full set of proteins in blood plasma), moved, and most had returned to baseline by 3 hours. Saliva told a different story. 601 proteins (31%) changed, and they stayed altered through the 24 hour mark. Urine was the quietest. 272 proteins (10%) shifted briefly and had fully resolved by 24 hours. Then came the replication test. In the older, mixed-sex cohort, 1,212 proteins (56% of the discovery set) responded again. About 74% moved in the same direction immediately after exercise, but that agreement dropped to 38% by 24 hours. Exercise type mattered less than you might expect. Aerobic, HIIT, and resistance sessions mostly changed how much a protein moved, not which direction it went. Only 2 proteins showed a true mode-by-time interaction (a pattern that depends on both the type of workout and when you measure it). For anyone choosing between steady cardio, intervals, and lifting, that is the practical takeaway. The core molecular response to a single session doesn't appear to depend on which one you pick. The dose you repeat matters more than the format. Put the two timing results together and a pattern seems to emerge. The first hours after a session look largely universal across age, sex, and workout type. By 24 hours, the responses diverge and become individual. That raises the obvious question of what the different fluids are doing, and why they follow such different clocks. The answer starts with the fluids themselves. First, each fluid has its own kinetics (the speed at which proteins rise and clear). Plasma acts like a highway with fast on-ramps and fast exits. Proteins spike, then get cleared. Saliva reflects local gland activity plus signals from the liver and immune cells, which is why it lingers. Urine mostly captures filtered leftovers from tissue turnover and the kidneys. That difference in timing sets up the next question: if the fluids run on separate clocks, are they still connected? Second, the fluids are talking to each other. The team measured how proteins in one fluid track with proteins in another across participants. Cross-fluid correlations rose about 1.7-fold immediately after exercise. Early hubs involved vesicle trafficking (small packages cells use to ship proteins) and immune activation. By 24 hours, the networks leaned toward growth factors and tissue repair. Third, the extracellular matrix (ECM, the scaffold of collagen and related proteins that holds tissues together) stayed central at every timepoint. The authors propose that matrix remodeling helps route exerkines to where they need to go. That is a hypothesis, not something they tested directly. Then they asked which proteins are likely to matter for disease. They used cis-pQTL colocalization, a genetic method that checks whether the same DNA variant drives both a protein's blood level and a disease risk. Proteins that responded consistently across age, sex, and exercise type were enriched for links to cardiometabolic disease. Mode-specific proteins mostly linked to physiological traits like body composition and kidney function. Of the secreted proteins with a genetic signal, 69 are targets of FDA-approved drugs. Some of the strongest signals showed up in saliva. 52 of 145 proteins in one sustained cluster were linked to type 2 diabetes risk. Why this matters: the proteins that move after one session are not random noise. They sit in the same biology that drives metabolic disease, and many are already druggable. That is a plausible route by which regular training lowers risk. This study doesn't prove it does. Finally, the team tested whether any of these proteins actually do something. A protein that rises after exercise may matter, or it may just be along for the ride. They used fruit flies (Drosophila), where muscle genes are easy to edit. In each experiment, they turned down one gene only in muscle, trained the flies for four days, and measured endurance. They tested nine proteins that humans release during exercise. Some were needed for the training benefit. With less PDGF, TNF, or FGF in muscle, flies lost their endurance gains, while untrained endurance didn't change. Others acted like a brake. With less myostatin (a protein that limits muscle growth), baseline endurance went up and training added nothing on top. With less GDF15 or BMP4, endurance rose in trained and untrained flies. Turning down INHBB even changed night sleep and fat storage, which suggests these signals reach beyond muscle. So at least some of these proteins help regulate adaptation. These are fly results, so they show the idea is plausible, not that the same thing happens in people. The limitations are real. There was no resting control group, so circadian rhythm, fasting, and fluid loss can't be fully separated from exercise. The authors controlled posture, fasting, and hydration, and many known exerkines reproduced. But it is still a design gap. The discovery group was 25 young men. Replication covered plasma only, so the saliva and urine findings haven't been replicated. Urine was sampled at just three timepoints. Most new exerkines were measured on one platform and need confirmation. The genetic colocalization is exploratory and doesn't prove causality. And flies are not mammals. The authors say a rodent model is the next step. This also isn't a training prescription. Nobody measured whether these proteins predict who gets healthier. Clinical use, such as biomarkers for exercise dose or drugs that mimic exercise, is likely years away. What it does give us is a map. The conserved response to a single session overlaps with the biology of diabetes, heart disease, and dementia. It also appears across ages, sexes, and workout types, which fits the idea that consistent movement matters more than the perfect format. No single drug is likely to recreate what exercise does across this many organs. The benefits of exercise aren't one signal from one tissue. They're a coordinated conversation across the whole body, and a single workout starts it.