Exercise Leaves Distinct Protein Fingerprints in Blood That Track Memory Gains in Older Adults

For decades, scientists have known that exercise sharpens the aging mind, but the molecular story behind that benefit has remained frustratingly opaque. Now a team of Canadian and French researchers has taken one of the most detailed looks yet at what happens in the blood when sedentary older adults start moving, and their findings suggest that different kinds of physical activity write different molecular messages into the bloodstream — messages that appear to be linked to improvements in episodic memory. The study, published in the journal GeroScience, used a high-throughput proteomic platform to track hundreds of inflammation-related proteins before and after a twelve-week training program, revealing both a shared signature of physical activity and a distinct aerobic-specific fingerprint.

The research emerged from a randomized, three-arm intervention trial conducted at the Research Center of the Montreal University Geriatric Hospital between 2015 and 2017. Healthy, physically inactive adults aged 60 to 79 were randomly assigned to one of three twelve-week programs: aerobic exercise on a recumbent bicycle, gross motor abilities training focused on balance, coordination and agility, or computerized cognitive training targeting executive functions. Each program involved sixty-minute supervised sessions three times a week. For this exploratory sub-analysis, the researchers focused on the 34 participants who had valid blood samples both before and after the intervention, profiling their plasma using the Olink Explore 384 Inflammation panel, a proximity extension assay capable of measuring 384 proteins with high sensitivity. After rigorous quality control, 356 proteins were retained for analysis.

The biological backdrop to the study is a phenomenon known as inflammaging — the chronic, low-grade inflammation that accumulates with age and is now considered a hallmark of biological aging. Senescent cells, which pile up in aging tissues, remain metabolically active and secrete a cocktail of pro-inflammatory molecules collectively called the senescence-associated secretory phenotype. This systemic inflammatory environment is a common feature of virtually all age-related chronic diseases, including the cognitive dysfunction that precedes neurodegenerative conditions. Exercise is thought to counteract this process by modulating cerebral blood flow, releasing neurotrophins, enhancing hippocampal neurogenesis and dampening neuroinflammation, but the specific circulating proteins involved have been difficult to pin down with traditional biomarker approaches that measure only a handful of molecules such as C-reactive protein or interleukin-6.

When the researchers compared pre- and post-intervention samples within each group, the picture was initially quiet. Only two proteins — PLXNA4, involved in axonal guidance and dendritic morphogenesis, and DAPP1, a participant in PI3K-mediated immune signaling — were significantly upregulated after training, and both were in the aerobic exercise group. No proteins changed significantly in the gross motor or cognitive training groups. The authors interpret this muted within-group response as evidence that systemic inflammation is relatively stable in healthy older adults, consistent with previous meta-analyses showing that peripheral immune markers are remarkably steady in healthy populations. The real signal emerged only when the researchers compared the pre-to-post changes between groups, a design that leverages each participant as their own control.

Those between-group contrasts proved striking. Aerobic exercise training produced a broader molecular modulation than any other intervention, with 57 proteins differentially abundant compared with the cognitive training group, while gross motor abilities training showed 25 such proteins against the same comparator. Crucially, 24 of the 25 proteins upregulated by gross motor training were also upregulated by aerobic training, revealing a common signature shared by both forms of physical activity. Pathway enrichment analysis of this shared signature pointed to MAPK cascade regulation, metabolic stress responses, immune receptor signaling including B cell receptor pathways, and cell growth regulation — processes the authors describe as consistent with a general adaptive response to repeated muscle contraction rather than pathological inflammation.

Beyond that common core, aerobic exercise carried a distinct molecular stamp of its own. The 33 proteins uniquely associated with the aerobic intervention were enriched for Toll-like receptor signaling — particularly TLR9, which recognizes mitochondrial DNA released during cellular stress, a phenomenon previously described following endurance exercise — as well as the calcineurin-NFAT pathway, long implicated in skeletal muscle fiber adaptation and vascular remodeling, and processes governing vascular smooth muscle cell differentiation. Proteins such as FGF2, EPO, TIMP3 and PLA2G4A, all with established roles in angiogenesis and responses to mechanical stress, featured in this aerobic-specific signature. Taken together, the authors suggest these pathways reflect an inflammatory adaptation to vascular and muscular remodeling that is characteristic of endurance training specifically.

The cognitive results added an intriguing layer. Across a nine-test battery, the type of intervention significantly influenced outcomes on five measures, including Stroop inhibition and switching, dual-task performance, and delayed recall on the Brief Visuospatial Memory Test-Revised. As expected, the cognitive training group outperformed both exercise groups on the executive function tasks, but the aerobic group showed a larger improvement in episodic memory than either comparator. When the researchers then asked whether changes in protein levels tracked changes in cognition, only one test yielded a significant association: delayed recall. A signature of 51 proteins was positively associated with improvements in visuospatial delayed recall, and enrichment analysis revealed overrepresentation of NF-κB signaling, PI3K/AKT signaling, antimicrobial humoral immune response and negative regulation of apoptosis — a profile the authors interpret as a regulated, adaptive neuroimmune state rather than maladaptive inflammation, consistent with pathways known to support synaptic plasticity, neuronal survival and memory consolidation.

Perhaps the most provocative finding concerns what did not explain the memory gains. Changes in peak oxygen uptake, the gold-standard measure of cardiorespiratory fitness, were positively correlated with memory improvement across the whole sample, but this association disappeared after adjusting for intervention group, suggesting it was driven by between-group differences rather than a direct within-subject effect. Moreover, the 51-protein memory signature showed no correlation whatsoever with fitness changes. This echoes previous reports that improvements in aerobic capacity alone are insufficient to account for exercise-induced memory benefits, and hints that other biological adaptations — possibly immune or vascular in nature — may carry the cognitive payload. The authors are careful to note that their study was not powered to exclude a fitness-related mechanism entirely.

Significant caveats temper these conclusions. The analysis involved a small and unequal subset of the original trial, with only six participants in the cognitive training arm, no non-intervention control group, and no preregistration. The absence of a passive control means time-dependent effects, regression to the mean and seasonal variation cannot be ruled out, and the multiple covariate adjustments relative to sample size raise the possibility of model overfitting. The participants were also healthy older adults without major comorbidities, limiting generalizability to clinical populations. The authors explicitly frame the findings as hypothesis-generating rather than confirmatory, and call for replication in larger, adequately powered cohorts with balanced group sizes and control conditions.

Even with those limitations, the study marks a meaningful step forward in the effort to decode how movement translates into mental sharpness late in life. By moving beyond single biomarkers to a 356-protein inflammatory panel, the researchers have shown that the molecular consequences of exercise are modality-specific, that a shared core response to physical activity exists alongside aerobic-unique pathways, and that circulating proteins may serve as readable indicators of the biological processes accompanying memory improvement. With 57 million people worldwide living with dementia and roughly 10 million new cases each year, understanding which forms of exercise produce which molecular signatures — and which of those signatures matter for the aging brain — could ultimately help design targeted interventions to preserve cognition and promote healthier aging.

Subject of Research: Plasma proteomic signatures of aerobic and motor exercise interventions and their association with episodic memory in older adults

Article Title: Common and distinct plasma proteomic signature of aerobic and motor physical activity interventions and their link to episodic memory

Article References: Mury, P., Jaulgey, K., Vrinceanu, T., Pothier, K., Berryman, N., Intzandt, B., Lussier, M., Vu, T.-T. M., Nigam, A., Bosquet, L., Karelis, A. D., Li, K. Z. H., & Bherer, L. (2026). Common and distinct plasma proteomic signature of aerobic and motor physical activity interventions and their link to episodic memory. GeroScience. https://doi.org/10.1007/s11357-026-02522-2

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02522-2

Keywords: proteomics, physical activity, aerobic exercise, inflammation, inflammaging, episodic memory, aging, cognitive function, biomarkers, older adults, Olink, GeroScience