Plasma proteomics identify novel biomarkers and dynamic patterns of biological aging

Abstract

Introduction: Plasma proteomics examines levels of thousands of proteins and has the potential to identify clinical biomarkers for healthy aging.

Objectives: This large proteomics study aims to identify clinical biomarkers for healthy aging and further explore potential mechanisms involved in aging.

Methods: This study analyzed data from 51,904 UK Biobank participants to explore the association between 2,923 plasma proteins and nine aging-related phenotypes, including PhenoAge, KDM-Biological Age, healthspan, parental lifespan, frailty, and longevity. Protein levels were measured using proteomics, and associations were assessed with a significance threshold of P < 1.90E-06. We utilized the DE-SWAN method to detect and measure the nonlinear alterations in plasma proteome during the process of biological aging. Mendelian randomization was applied to assess causal relationships, and a PheWAS explored the broader health impacts of these proteins.

Results: We identified 227 proteins significantly associated with aging (P < 1.90E-06), with the pathway of inflammation and regeneration being notably implicated. Our findings revealed fluctuating patterns in the plasma proteome during biological aging in middle-aged adults, pinpointing specific peaks of biological age-related changes at 41, 60, and 67 years, alongside distinct age-related protein change patterns across various organs. Furthermore, mendelian randomization further supported the causal association between plasma levels of CXCL13, DPY30, FURIN, IGFBP4, SHISA5, and aging, underscoring the significance of these drug targets. These five proteins have broad-ranging effects. The PheWAS analysis of proteins associated with aging highlighted their crucial roles in vital biological processes, particularly in overall mortality, health maintenance, and cardiovascular health. Moreover, proteins can serve as mediators in healthy lifestyle and aging processes.

Conclusion: These significant discoveries underscore the importance of monitoring and intervening in the aging process at critical periods, alongside identifying potential biomarkers and therapeutic targets for age-related disorders within the plasma proteomic landscape, thus offering valuable insights into healthy aging.

Keywords: Biological age; Biomarkers; Health aging; Nonlinear change; Proteomics.

Graphical abstract

Fig. 1

Relationship between health aging and plasma proteins. A. Scatter plots show the associations between 9 aging-related traits and 2,923 proteins. The Cox regression model was applied to examine the association between protein levels and healthspan. Linear regression models were applied to examine the association between protein levels and PhenoAge, PhenoAge acceleration, KDM-BA, KDM-BA acceleration, telomere length, and parental life span. Logistic regression models were applied to examine the association between protein levels and frailty and longevity. The displayed p-values were two-sided and adjusted for multiple comparisons. Proteins above the horizontal dotted black line were significantly linked to aging-related traits following Bonferroni corrections (P < 1.90 × 10–6). B. Plots show the number of each category of proteins significantly associated with aging-related traits, including four categories: cardiometabolic, inflammation, neurology, and oncology. C. Results of Mendelian randomization analysis on associations of plasma proteins with healthy aging. Forest plot showing causal effects of 5 MR-identified proteins on healthy aging. The boxes represent odds ratio (OR) values, and the horizontal lines represent 95 % confidence intervals.

Fig. 2

Biological function of health aging-associated proteins. A. Plot demonstrates tissue-specific expression patterns of health aging-related proteins. Analysis was conducted using GENE2FUNC in Functional Mapping and Annotation (FUMA), based on the GTEx v8 database comprising 54 tissue types. B. Results of the pathway enrichment analysis. The x-axis represents the −log10 of the P value for each term, indicating the statistical significance after FDR correction. Different terms are listed on the y-axis, with their sources distinguished by various colors. C. Results of the transcription factors enrichment analysis. The transcription factors that ranked among the top 5 in significant enrichment after FDR correction are illustrated. Abbreviations: KEGG, Kyoto Encyclopedia of Genes and Genomes pathways.

Fig. 3

Plasma proteomes trajectories during biological aging. A. The study examined plasma protein trajectories throughout the aging process. Utilizing z-score normalization, trajectories were estimated for 227 plasma proteins using the LOESS method. B. A heatmap visually displayed the trajectories of plasma proteins during the aging. C. The heatmap specifically highlighted the trajectories of 5 aging-related proteins during the aging. D. Clustering analysis identified four distinct groups of plasma protein trajectories during aging. The average trajectory of each cluster was visually emphasized with a thicker line, and the number of proteins within each cluster was annotated for reference.

Fig. 4

Plasma protein waves during biological aging. A. The upper portion illustrates the variations in the number of plasma proteins with differential expression during the aging process. Notably, three peaks were observed at biological age 41, 60, and 67, indicating distinct phases or stages of aging. B. Overlaps between waves of biological age proteins. C. The heatmap illustrated the importance of 5 aging-related proteins at biological ages 41, 60, and 67. D. Identify the key biological processes associated with each protein wave as generated by KEGG. E. The variations in plasma protein expression during the aging process differ across various organ tissues. Except for the lungs, three peaks were observed in other organs at biological ages, indicating distinct phases or stages of aging. Notably, the first peak in the lungs appeared latest, at the age of 60.

Fig. 5

Phenome-wide association analysis of aging-associated proteins. A-B. phenotype-wide association analysis between 5 key aging-related proteins and organ phenotypes, diseases, and mortality. The y-axis indicates the − log10 of the P values for each association, and the x-axis represents different phenotype categories. The P values shown are two-sided and adjusted for multiple testing. Grey line in each figure is Bonferroni 0.05 correction threshold. C. The association between LE8 and aging. D. The bar chart displays the proportion of mediating proteins and incident outcomes for each exposure. This indicates the proportion of the impact of modifiable risk factors on aging mediated through protein intermediates. Detailed median proportion significantly mediated by each protein across health aging is provided.