Cold Exposure Reshapes Human Blood Proteins Toward Heart Protection and Slower Aging
A first-of-its-kind proteomics study finds acute cold exposure shifts 185 circulating proteins in a cardioprotective, anti-aging direction across independent cohorts.
Summary
Researchers at Rockefeller University and Beth Israel Deaconess measured nearly 7,200 blood proteins before and after controlled cold exposure in healthy adults. They found 185 proteins changed consistently across two independent cohorts, with most decreasing — including Leptin, FABP4, and Kallikrein proteases linked to inflammation and cancer. Proteins associated with aging, cardiovascular disease, and metabolic dysfunction shifted favorably. The cold-responsive proteins were disproportionately sourced from adipose and smooth muscle tissue, and many overlapped with the secretome of human brown adipocytes grown in the lab. The findings suggest cold therapy may work partly through blood-borne signaling molecules, not just calorie burning, offering new molecular targets for longevity and cardiometabolic medicine.
Detailed Summary
Cold exposure — through ice baths, cryotherapy, or cold vests — is widely promoted for health benefits, yet the molecular mechanisms remain poorly understood. This preprint from Rockefeller University, Beth Israel Deaconess, and the NIH provides the most comprehensive circulating proteome analysis of acute cold exposure in humans to date, using two independent cohorts and two high-sensitivity protein measurement platforms to identify a reproducible molecular signature of cooling.
The discovery cohort enrolled 20 healthy young adults (10 male, 10 female; ages 21–28; BMI 18.5–25) who underwent 3 hours of personalized cooling via a water-perfused vest set 2°C above their individual shivering threshold, with blood drawn after an overnight fast. A separate fasting-only control protocol allowed the team to distinguish cooling-specific effects from fasting effects. The validation cohort included 18 subjects (12 male, 6 female; mean age 26; mean BMI 24.5) with BAT confirmed by ¹⁸FDG-PET/CT scans, cooled for 1–2 hours. Together, the two cohorts yielded measurements of 7,172–7,198 unique circulating proteins using SOMAscan and Olink platforms.
Of 225 proteins significantly altered by cooling in the discovery cohort (FDR<0.05), 185 (81.5%) replicated in the validation cohort with the same directionality — a striking concordance given different cooling durations and protocols. The majority of changes were decreases: 165 proteins fell and only 20 rose with cooling. Top cold-suppressed proteins included KLK8 (β=−1.42, p=2.53×10⁻⁸), LY6D (β=−1.36, p=1.66×10⁻⁸), SPINK6 (β=−1.24, p=2.17×10⁻⁸), Leptin (β=−0.37, p=1.19×10⁻⁸), and FABP4 (β=−1.07, p=1.23×10⁻¹¹). Sixteen Kallikrein serine proteases — enzymes linked to inflammation, coagulation, and cancer — were consistently suppressed by cooling but not by fasting alone, suggesting a cold-specific mechanism. Top upregulated proteins included APOC1, NPTXR, FGA, and ARSK.
Tissue mapping via the Human Protein Atlas revealed that adipose-derived proteins showed the second-highest rate of cold responsiveness (49% of 144 adipose-mapped proteins changed), while smooth muscle proteins showed the highest (55% of 42 proteins). Skeletal muscle and liver proteins changed at lower rates (35% and 33%, respectively). Many cold-responsive proteins overlapped with the secretome of human iPSC-derived brown adipocytes studied in vitro, supporting the hypothesis that BAT activation contributes to the circulating proteomic shift. Pathway analyses linked the cold-responsive proteome to reduced biological aging markers, improved cardiovascular risk profiles, and favorable metabolic changes — consistent with prior epidemiological findings that BAT-positive individuals have 56% lower odds of type 2 diabetes independent of BMI.
The study also performed phenome-wide association analyses (PheWAS) connecting the cold-responsive proteins to clinical traits, reinforcing the cardiometabolic and anti-aging relevance of the identified signature. ELISA validations of selected proteins confirmed concordance with aptamer and antibody platform data (p<0.01 for all validated proteins). Cystatin C levels — a proxy for kidney filtration rate — were associated with the overall leftward skew in protein levels, suggesting cooling may accelerate plasma protein clearance or dilution, a nuance the authors addressed through residualization analyses. This is a preprint and has not yet undergone peer review, and the cohorts are small, young, and healthy, limiting generalizability to older or metabolically compromised populations.
Key Findings
- 185 of 225 cold-regulated proteins (81.5%) replicated across two independent cohorts with identical directionality, establishing a high-confidence cold-exposure proteome signature
- 165 proteins decreased and only 20 increased with acute cooling, with top suppressions including KLK8 (β=−1.42, p=2.53×10⁻⁸), FABP4 (β=−1.07, p=1.23×10⁻¹¹), and Leptin (β=−0.37, p=1.19×10⁻⁸)
- 16 Kallikrein serine proteases linked to inflammation, coagulation, and cancer were suppressed by cooling but not by fasting alone, indicating a cold-specific mechanism
- Adipose-derived proteins showed 49% cold-responsiveness rate (71 of 144 proteins), and smooth muscle proteins showed the highest rate at 55% (23 of 42 proteins)
- Cold-responsive proteins significantly overlapped with the secretome of human iPSC-derived brown adipocytes in vitro, implicating BAT activation as a key source
- PheWAS analyses linked the cold-responsive proteome to reduced biological aging markers and improved cardiovascular and metabolic disease risk profiles
- 7,198 unique circulating proteins were quantified using combined SOMAscan and Olink platforms — the broadest proteomic cold-exposure analysis in humans to date
Methodology
Two cohorts of healthy young adults (discovery: n=20, ages 21–28, BMI 18.5–25; validation: n=18, BAT confirmed by ¹⁸FDG-PET/CT, mean age 26, mean BMI 24.5) underwent controlled cooling via water-perfused vests, with blood drawn at room temperature and post-cooling. A fasting-only control arm in the discovery cohort separated cooling-specific from fasting-specific protein changes. Proteomics used SOMAscan and Olink platforms measuring up to 7,198 unique proteins; statistical significance was assessed by linear mixed models with FDR correction, and ELISA validations confirmed key findings.
Study Limitations
The study cohorts are small (n=20 and n=18) and consist exclusively of young, healthy adults with normal BMI, limiting generalizability to older, obese, or metabolically diseased populations. As a preprint, the findings have not yet undergone peer review. The authors acknowledge that the overall decrease in circulating proteins may partly reflect cold-induced changes in kidney filtration or plasma volume rather than purely secretory changes, which they attempted to address through Cystatin C residualization.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.