Measuring the proteins circulating in mouse blood has long been a frustrating exercise in compromise. The most informative assays demand milliliters of serum, yet a mouse offers only a few drops, and longitudinal studies that draw blood repeatedly leave researchers with volumes too small for conventional deep proteomics. A team reporting in Nature Aging has now built a workflow that squeezes comprehensive protein measurement out of as little as 20 microliters of serum, roughly a hundredth of a milliliter, and used it to chart how the blood proteome changes across the mouse lifespan.
The method, developed by Amit K. Dey, Nathan Basisty and colleagues at the National Institute on Aging with the Study of Longitudinal Aging in Mice investigator group, pairs automated nanoparticle-based protein enrichment with data-independent acquisition mass spectrometry. Each serum sample is incubated with five chemically distinct nanoparticles built on superparamagnetic iron oxide cores. Proteins bind to these particles in characteristic coronas, and because each particle surface attracts a different slice of the proteome, low-abundance proteins that would otherwise drown beneath albumin and other dominant species are pulled into measurable range. The technique works by dynamic range compression: instead of trying to detect every protein directly in an undiluted digest, the workflow partitions the sample across five parallel enrichment channels.
The gain in coverage was substantial. In a feasibility test with pooled mouse serum, nanoparticle processing increased protein depth two- to four-fold compared with standard digests of unprocessed serum. When the team applied the full workflow to 30 C57BL/6J mice from the SLAM cohort, aged 12, 24 and 30 months with 10 mice per age and half female in each group, they quantified 3,992 protein groups from 20 microliters of serum per animal, a sixfold improvement over the standard neat workflow. Across all 30 mice the method detected 28,976 unique peptides, with individual nanoparticle fractions contributing between roughly 9,600 and 24,500 peptides each. Median coefficients of variation stayed below 10 percent, indicating that the enrichment is not just deep but reproducible.
Depth means little without quantitative fidelity, so the researchers ran a benchmarking exercise using mixtures of human and bovine plasma prepared at known ratios, from 5 percent to 100 percent bovine. After nanoparticle processing, the measured peptide abundance ratios tracked the expected values closely: the median ratios of the most divergent mixtures fell within 0.01 of what was expected, and the intermediate mixtures landed within their predicted interquartile ranges. The team also compared two computational strategies for rolling peptide measurements up to protein values, MaxRep and MaxLFQ, and found the two correlated strongly, with median protein-level coefficients of variation of 0.0567 and 0.0613 respectively. They adopted MaxRep for the primary analyses on the strength of its marginally better reproducibility.
One category of biologically important proteins stubbornly resists mass spectrometry: cytokines. These small signaling molecules sit at the very bottom of the blood protein dynamic range and carry too few distinctive peptide fragments to detect reliably. Interleukin-6, a workhorse inflammatory marker and a classic senescence-associated secreted protein, was conspicuously absent from the mass spectrometry data. To close that gap, the researchers spent a single additional microliter of serum per mouse on a targeted 48-plex proximity extension assay, which quantified 43 cytokines that mass spectrometry had missed. Only four proteins appeared in both datasets, underscoring how complementary the two platforms are. Where overlap existed, the measurements agreed: interleukin-16 levels correlated significantly between the two methods, with a correlation of 0.7455.
The aging study itself revealed extensive remodeling of the circulating proteome. Linear modeling adjusted for sex identified 73 age-associated proteins among the 2,662 proteins that met quality filters, with 54 increasing and 19 decreasing as the mice moved from young adulthood through old age into geriatric age. Among the strongest risers were TXNDC5, MYL13, ARPC5, LGMN and LMAN1. Myosin light chain 3, a core muscle component known to rise in serum during muscle wasting, featured among the most strongly age-linked proteins, while paraoxonase 1, an atheroprotective enzyme that declines with age, moved in the opposite direction. Gene set enrichment analysis pointed to spermatogenesis and immunity-related pathways changing between 12 and 24 months, and metabolic processes becoming prominent between 24 and 30 months, hinting at stage-specific biology in the oldest animals.
Sex left as large a fingerprint on the blood proteome as age did. The modeling identified 20 proteins elevated in males and 61 elevated in females by mass spectrometry, and the proximity extension assay added one more, interleukin-7, higher in females. Male-elevated proteins were enriched for plasma membrane components, while female-elevated proteins clustered around peptidase activity. When the team compared their sex-associated proteins against prior mouse proteomics studies, including an extensively fractionated serum analysis in an Alzheimer disease model and a targeted plasma survey across 30 knockout strains, all but 2 of 32 overlapping proteins showed the same directional sex association, a strong external consistency check for the new workflow.
Serum from aged mice also carried a heavier senescence signature. Compared with the fewer than 50 senescence-associated proteins detected in unprocessed serum using the SenSig reference set, nanoparticle processing surfaced 136 SenSig proteins in processed samples and 209 across all aging study samples. The cytokine panel told a consistent story: ten cytokines rose significantly and linearly with age, including recognized senescence-associated secreted factors such as interleukin-6, TNF, CSF3, CSF1 and CCL2, while interleukin-21 and CCL22 declined. CCL12 showed the most significant age-related change. Taken together, the authors argue, these patterns support an age-related increase in senescence-driven immune cell activation in circulation, though the cross-sectional design means the data capture population-level differences between age groups rather than trajectories within individual animals.
The translational payoff came from comparing the mouse findings with human plasma proteomics from the Baltimore Longitudinal Study of Aging, which profiled 1,275 participants with a median age of 70 using the SomaScan 7k assay. Of the 1,742 proteins quantified in both datasets, 18 were significantly and positively associated with age in both species, representing 5 percent of the shared proteins but fully half of the mouse age-associated set. Overrepresentation analysis showed these conserved proteins clustered in cellular component categories including endoplasmic reticulum, extracellular matrix and lysosomal proteins. Five of them, COL6A3, LBP, PIGR, IGFBP1 and HPX, were also significantly associated with a 44-component frailty index in the human cohort. COL6A3, the collagen VI alpha 3 chain also known as endotrophin and a marker of fibrotic aging, was by far the strongest frailty association, with an adjusted P value of 5.069 × 10−16, and it ranks among the top five proteins linked to prefrailty and frailty in the UK Biobank.
The authors are careful about what the study does and does not establish. The cohort was modest, with 30 mice and 10 per age group, limiting statistical power, and because different mice were sampled at each age rather than the same animals followed over time, the age-associated patterns reflect between-group differences rather than within-individual aging trajectories. The findings are framed as foundational and hypothesis-generating rather than mechanistic or causal, and future phases of SLAM with larger samples and true longitudinal sampling are planned to test whether these circulating signatures hold up within individuals. Even so, the methodological achievement stands on its own: a scalable, automated pipeline that turns a 25-microliter serum budget into nearly 4,000 quantified proteins plus 43 cytokines, compatible with the volume constraints that have kept most mouse aging studies proteome-free, and already pointing toward a short list of conserved blood proteins, led by endotrophin, that track functional decline in both species.
Subject of Research: A nanoparticle enrichment and mass spectrometry workflow for deep proteomic profiling of low-volume mouse serum in aging research
Article Title: Deep and quantitative proteomic profiling of low-volume mouse serum in a cross-sectional analysis of aging
Article References: Dey, A. K., Olinger, B., Boroumand, M., Fernandez, M. E., Study of Longitudinal Aging in Mice (SLAM) Investigators, de Cabo, R., Camandola, S., Price, N. L., de Cabo, R., & Basisty, N. (2026). Deep and quantitative proteomic profiling of low-volume mouse serum in a cross-sectional analysis of aging. Nature Aging. https://doi.org/10.1038/s43587-026-01242-8
Image Credits: AI Generated
DOI: 10.1038/s43587-026-01242-8
Keywords: proteomics, mass spectrometry, nanoparticles, aging biomarkers, mouse serum, cytokines, senescence, frailty, Baltimore Longitudinal Study of Aging, COL6A3, data-independent acquisition, SLAM cohort
News Source: Beatrice Stafford. (November 5, 2026). Nanoparticle workflow reads aging signatures from just 20 microliters of mouse serum. Scienmag.
